Split (1)

train · 19.4M rows

text stringlengths 9 94.9k
Help! A giant red sea snake attacked a sailing ship! Hopefully it goes well!
A detail of an ancient nautical chart rests in a square brass medallion with a thick glass cabochon.
The medallion is hanging with a thick red glass fish, which ensures that the medallion is always in the center of the necklace, and some turquoise beads on a brass chain.
This jewel is present only once in this jewelry store!
This Product was added to our catalogue on Monday 29 August, 2016.
In addition to trainings, exercises, and the sharing of tools and resources, AzCHER-Central provides education to its members and the community at large to help elevate the collective knowledge of emergency preparedness and emergency management.
One such example of the coalition’s education efforts is found in our partnership with the Center for Public Health Law and Policy at ASU’s Sandra Day O’Connor College of Law. Professor James Hodge has prepared and recorded the following 2018 video to assist emergency managers and healthcare workers in understanding the legal and ethical issues surrounding emergency preparedness in Arizona. A PDF of the notes accompanying the presentation can be downloaded here.
If you are studying or traveling in Germany, you learn some German. In the event of an emergency, you need to be able to communicate and understand effectively. You need to be able to ask for help if injured, and ask for legal protection if your rights are in jeopardy. If you have a specific health condition, a special need, or if you are allergic to any medication, know exactly how to say so in German. Regardless of your language proficiency level, there are a few basic words, phrases, and questions that you should be able to pronounce fluently. You should fill out our Words To Know Sheet and take it with you for reference.
Each student should develop his or her own personal list of essential words and phrases to know. You may encounter many other helpful and important phrases you would like to include on your own list. For instance, you may want to look up specific phrases pertaining to Germany. Use our list as a guide, or starting point; then add your own additional Germany–specific phrases. In addition, you should also know how to dial a country's 24–hour emergency phone number (equivalent to a 9–1–1 system).
You may find it helpful to print and cut out our Service Icons Card that you can carry with you. The card contains a series of simple images that represent services you may require—services like a telephone, post–office, taxi, and hospital. If you don't remember, or don't know, how to say a certain word in German, you can point to the picture on the card that represents that word. For example, if you need to find a telephone, you can show someone the picture of a telephone on the card and they can assist you. It is better to already be able to comfortably communicate in German when studying in Germany. However, you should keep this card with you just in case you forget how to say a certain word, or, in the case of injury, you become physically unable to verbally communicate.
Russia's major companies budgeted a total of $20 billion for innovative programs in 2013, Russian Prime Minister Dmitry Medvedev said.
"I've just looked at the figures, only major companies allocated around 550 billion rubles for programs of innovative re-organization and re-equipment in 2013. This is about $20 billion," Medvedev said at the conference Dialog on the Future in Skolkovo.
The Russian prime minister said that these were "not fantastic figures but this is quite serious money already."
State institutions - venture structure, Rusnano corporation - are involved in innovative processes in Russia and are participating actively, Medvedev said.
Is used in transporting hazardous material under 49 U.S.C. 5103 and transported in a quantity requiring placarding under 49 CFR, subtitle B, chapter I, subchapter C.
I have monocular vision. I live in Florida and my CDL is restricted to intrastate only. Am I required to have an exemption or waiver?
I drive a tow truck in Florida. Do I need a CDL? Which vehicle group does it fall under?
If the Gross Combination Weight Rating (GCWR) of the tow truck and its towed vehicle is 26,001 pounds or more, and the towed vehicle alone exceeds 10,000 pounds Gross Vehicle Weight Rating (GVWR), then the driver needs a Group A CDL.
If the Gross Vehicle Weight Rating (GVWR) of the tow truck alone is 26,001 pounds or more, and the driver either (a) drives the tow truck without a vehicle in tow, or (b) drives the tow truck with a towed vehicle of 10,000 pounds or less Gross Vehicle Weight Rating (GVWR), then the driver needs a Group B CDL.
If neither of the above apply, and the driver tows a vehicle required to be placarded for hazardous materials on a ‘‘subsequent move,’’ i.e. after the initial movement of the disabled vehicle to the nearest storage or repair facility, then the driver needs a Group C CDL.
Can your office issue my medical certificate?
Yes. As of May 2014, all U.S.A. DOT medical examinations must be conducted by a Federal Motor Carrier Certified Medical Examiner. Dr. Green is a Federal DOT Certified Medical Examiner.
You can apply for a vision exemption here or a diabetes exemption here.
This product is an elephant-shaped smartphone stand.
It has been designed to help the endangered African Baby elephant.
We will donate 1% of the proceeds to the Baby Elephant Conservation Group.
Upgrading Our EMR Netsmart. Backing up the database and creating a sandbox server to text and try the new updates.
Took EMR Netsmart software from 17.0 to 18.4. Built a sandbox server (server 2012) backed up patient database and transferred it to the sandbox. Renewed SSLs.
Beyond Darwin: revealing culture-specificities in the temporal dynamics of 4D facial expressions.
Since Darwin’s seminal work on the evolutionary and biological origins of facial expressions, the Universality Hypothesis maintains that all humans express six basic emotions - "happy," "surprise," "fear," "disgust," "anger" and "sad" - using the same set of distinct facial movements. Testing this hypothesis directly, we used a novel platform to generate random 3-dimensional facial movements, which observers perceive as expressive when correlating with their mental representations. Fifteen Western Caucasian (WC) and 15 East Asian (EA) observers each categorized 4,800 (same and other race) animations according to the six basic emotions (or "don’t know") and by intensity ("very low" to "very high." Figure S1, Panel A). We then reverse correlated the random facial movements with the emotion responses they elicited, thus computing 180 models per culture (15 observers x 6 emotions x 2 race of face). Each model comprised a 41-dimensional vector coding the facial muscle composition and temporal dynamics. The Universality Hypothesis predicts that these models will form six distinct clusters (one per emotion) in each culture and show similar signaling of emotional intensity across cultures. We show cultural divergence on both counts: Cluster analysis of the models in each culture revealed that WC models optimally form 6 distinct and emotionally homogenous clusters as predicted (Levenson 2011), whereas EA models overlap between emotion categories, with little categorical structure (Figure S1, Panels B-C). Cross-cultural comparison of emotional intensity signaling across time (i.e., co-variation of facial movements and intensity) revealed further cultural differences. Whereas WC models signal emotional intensity with distributed face regions, EA models showed early signaling with the eyes, as mirrored by popular culture EA emoticons -- (^.^) "happy" and (O.O) "surprise." Here, we refute the Universality Hypothesis and raise the question: if the 6 basic emotions are not universal, which emotions are basic in different cultures?
Meeting abstract presented at VSS 2012.
Don't let the border stop you from experiencing all of Niagara Falls' glory! On this deluxe 6-hour tour, visit top Niagara Falls attractions on both the Ontario side and the New York side. Take a ride on the famous 'Maid of the Mist' boat, view the falls from atop Skylon Tower, feel the falls’ thunder at the Cave of the Winds and much more. All attraction fees, admission fees, hotel transfers and road tolls are included!
Im just wondering, i am in 6th division and currently 4th best team. Most of the weaker ones are like 10-15 total skill point lower, does that affect my experience gain from games?
Or is it the progress no matter of the opponent?
No, that doesn't affect the experience gain at all. The two things, that impact the gaining of experience is the time on ice and his age. The younger a player is, the more experience he gains.
Thank you for your answer, it helped me a lot!
Sorry for the following offtopic: Can you tell me how did you do that signature with your team information?
You log into your account and go to office, then to settings (the crossed hammer and wrench, I always overlooked it back in the day when someone told me to go to the settings ), then you see on the right side "Signature", it's the second last option. There you copy the BB-Code, then you go to the forum, click on profile, then signature and there you insert the code you copied before.
I'm not 100% sure you already can create a custom signature in the forum, it's linked to the amount of posts you created, but I believe you should have enough.
If Others, do state so.
Click here after submitting the form to pay now!
All payment to be payable to TRADEVSA SYSTEM SDN BHD, Maybank 514178647113. Send a copy of the payment slip to Zac at: +60-10-266-9761.
IMPORTANT: All payments will be made in Ringgit Malaysia. No refunds are allowed.
Traffic lights can be tricky in a real emergency. If the light is red, emergency vehicles lose precious time moving slowly and carefully through the intersection. But a pilot program and partnership between the City of Dunwoody and DeKalb County could make a big difference.
The Glance Priority and Preemption System by Applied Information is designed to give fire trucks the green light when responding to a 911 call. It makes it easier and safer for emergency vehicles by bringing all other traffic safely to a halt.
Special preemption equipment has been installed in two trucks at DeKalb County Fire Station 18 on Barclay Drive behind Peachtree Charter Middle School. Corresponding equipment has also been placed inside the signal cabinets at two intersections: Peeler Road at North Peachtree Road and Tilly Mill Road at North Peachtree Road.
The system works by letting the trucks and signals essentially talk to one another using GPS and cellular technology. As a fire truck on an emergency call approaches the intersection, the equipment on the truck sends a message to the lights that it’s coming.
If the light is red, the signal will cycle through and turn green in time to allow the truck to move safely through the intersection without slowing or waiting.
“We are pleased that DeKalb County Fire Rescue has partnered with the City of Dunwoody to enhance safety at these busy intersections and reduce emergency response time,” said Dunwoody Public Works Director Michael Smith.
The City of Dunwoody paid for the signal cabinet equipment, which costs about $5,000 per signal. DeKalb County picked up the cost of the truck equipment, which runs about $2,500 per vehicle.
Dunwoody is looking to install the equipment at a couple of other signals in the Perimeter area soon. The City will analyze the impact of this pilot program in exploring whether it can be expanded further.
Great post, Doug. Thanks. What is the old grocery today asks the wayward Alamedan in L.A.
The old grocery has been transformed into a residence. Check out the Redfin gallery (see link in the text) for a visual tour of the place.
Wow…wish I’d bought it…we could just walk on over…plus the art scene. Very nice.
Methods, systems and devices are implemented in connection with measuring subcutaneous fat and loin depth in a portion of muscle tissue. Consistent with one such method a probe is presented to the portion of muscle tissue. The probe produces a response-provoking signal in the muscle tissue used to determine the fat and loin depth in the portion of muscle tissue.
This patent document claims the benefit, under 35 U.S.C. §119(e), of U.S. Provisional Patent Application Ser. No. 61/050,542 filed on May 5, 2008, and entitled “SYSTEMS, METHODS AND DEVICES FOR USE IN ASSESSING FAT AND MUSCLE DEPTH;” this patent document and the Appendices filed in the underlying provisional application are fully incorporated herein by reference.
Subject matter described in this document is based upon work supported by the Cooperative State Research, Education, and Extension Service, U.S. Department of Agriculture, under Agreement Nos. 2006-33610-16761 and 2007-33610-18441. The U.S. government has certain rights to this invention.
The present invention relates to systems and methods for inspecting and measuring muscle tissue parameters related to fat and lean content and quality of meat.
There are several attributes of meat quality that relate to palatability and consumer eating satisfaction. Assessments of such qualities can be useful for a variety of food animals. Such assessments can also be useful in both live animals and animal carcasses. For example, carcass weight, backfat and/or loin depth measurements may be used to determine the value of pork. Fat-free lean in swine carcasses may be predicted from fat depth and loin depth measurements.
Current fat and depth measurement processes employ manual methods, optical probes, and cross-sectional scanning, and involve highly trained technicians performing time-consuming and labor-intensive work to achieve accurate measurements. Thus, a significant challenge is presented with measuring fat and loin depths in a packing plant environment at line speeds. As an example, with many plants running their chain speed at 1200 carcasses per hour, a carcass would be measured in less than 3 seconds if the carcass is going to be measured during the packing process. In addition, pork carcasses are not routinely split anywhere along the loin that would expose the internal tissue for either a subjective or quantitative measure of fat and loin depth.
The present invention is directed to systems and methods for determining quantitative measure of fat and loin depths of muscle tissue, for example, pork and beef, including from hot carcasses and live animals. These and other aspects of the present invention are exemplified in a number of illustrated implementations and applications, some of which are shown in the figures and characterized in the claims section that follows.
Consistent with an embodiment of the present invention, a method is implemented for automatically measuring fat and loin depth. An ultrasonic probe is presented to the carcass. The probe produces a response-provoking signal in the meat. A resulting signal is used to determine various tissue boundaries such as the skin-fat boundary, fat-muscle boundary, and rib-muscle boundary, so that fat and loin depth can be measured.
FIG. 7 shows a block diagram of an example intercostales muscles boundary automation algorithm, consistent with an example embodiment of the present invention.
The present invention is believed to be useful for inspecting and measuring muscle tissue parameters, such as fat and loin depth. The muscle tissue can originate from any number of different food animals and the inspection and measuring can be obtained from live animals or animal carcasses. A specific embodiment of the present invention facilitates measurement of fat and loin depth of a pork carcass. Unless otherwise stated, the term animal references to either a live animal or an animal carcass. While the present invention is not necessarily limited to such applications, various aspects of the invention may be appreciated through a discussion of various examples using this context.
An embodiment of the present invention is directed toward a noninvasive mechanism for determining fat and loin depth of muscle tissue, such as muscle tissue from live pork or beef animals, or pork or beef carcasses. Ultrasound imaging is used to capture internal images of the muscle tissue. An image processor processes the images using algorithms specifically selected and/or tailored to use with the particular muscle tissue (e.g., the type of food animal or whether live or dead) to determine the fat and loin depth in a carcass processing line. An embodiment of the present invention can include the step of using the fat depth and muscle depth to select livestock for breeding, and rate the portion based on quality criteria, to sort livestock.
Specific embodiments of the present invention are directed toward facilitating the determination of fat and loin depth in a meat processing line. Devices, methods and systems facilitate fat and loin depth determinations at speeds and accuracy levels that are particularly useful for use on a processing line. Various aspects include, for example, streaming image capture, image selection criterion, specifically tailored algorithms and/or facilitating proper contact between the carcasses and a probe.
An embodiment of the present invention is directed toward a noninvasive system for measuring subcutaneous fat depth and muscle depth in the longissimus dorsi muscle of hot carcasses, and using such measurements to aid in determining other muscle tissue characteristics, such as percentage intramuscular fat (IMF). The measurements are made real-time, for example, on carcasses that are moving on a transport rail at a nearly constant rate of 1,200 carcasses per hour. Measurements are made from live video-streaming ultrasound images as the carcasses move past a scanning station. The scanning station can be fully automated, manual or a combination thereof.
By way of example, FIG. 1 illustrates a system for use in inspecting and measuring muscle tissue parameters in carcasses, according to a specific embodiment of the present invention. Probe 102 communicatively connects to processing block 104 using probe input/output (I/O) 110, 112. This connection can be implemented using, for example, a wired connection, wireless connections or a removable storage medium. Wired connections can be implemented using any suitable (e.g., bandwidth and reliability) protocol including, but not limited to, universal serial bus (USB), IEEE 1394 and Ethernet. In a specific instance, the probe is connected using a data-carrying cable (e.g., electrical or optical). In another instance, the probe is integrated into a single device that includes the processing block 104. Wired connections can also be implemented using a more temporary connection, such as a removable data storage device or a cradle for placement of the probe. Wireless connections for non-ultrasound communications can be implemented using an assortment of different techniques and protocols including, but not limited to, 802.11x or ultra-wideband (UMB).
With respect to image acquisition conditions, it has been observed that the quality of ultrasonic images acquired from animal carcasses can be affected by the pressure applied between the probe transducer and the carcass. Thus, in reference to FIG. 1, the probe 102 optionally includes one or more pressure sensors such as load cells 116A and 116B. Information from the pressure sensors may be used by an optional image filter 118 within the probe 102 to decide whether to capture and transmit images to the processing block 104. In other embodiments, the pressure data is transmitted to the processing block 104 for analysis, at which point the images may be recorded using video capture 128 and/or buffer 122 and retained for further analysis or discarded based on the pressure readings. In another example, the processing block 104 analyzes the pressure data and in response determines whether or not to activate the ultrasound transducer. Feedback signals may be provided to control further image acquisition by the probe and/or to provide an operation status indication (e.g., yellow light for non-acquisition stand-by mode when the probe is not applied or insufficient pressure is applied, red light for non-acquisition based on too much pressure or unbalanced pressure, and green light for ultrasonic activation and image acquisition due to proper application of the probe).
According to specific embodiments of the present invention, processing parameters 120 used by the algorithms for determining muscle tissue characteristics can be dynamically adjusted for each carcass. For example, each carcass has a specific makeup with regards to the tissue depths of various tissue types. These differences can affect the captured image data as, for example, different tissue types can exhibit different sound propagation properties. Tissue types that can be monitored for dynamic adjustments include, but are not limited to, subcutaneous fat, muscle (loin), skin and bones. In a specific instance, the subcutaneous fat depth and loin depth within a region of interest are determined. These determined depths may then be used as further parameters in algorithms for determining other muscle tissue characteristics, such as IMF.
Ultrasound device calibration can be particularly useful for maintaining consistency between measurements where, for example, components of the device are replaced or operation parameters change (e.g., changes over time due to use or due to temperature variations). One mechanism for calibration involves the use of a default device that is already calibrated. Measurements are taken for each device and the parameters for the device under calibration are modified so that the results coincide with the results of the default device. Another mechanism involves the use of a known item from which fat and loin depth measurements are taken. The item could be one or more carcasses.
By way of example, FIG. 2 illustrates a meat packing plant environment where hot carcasses, such as carcass 294, are conveyed along an overhead conveyor system 292 in a direction indicated by the arrow. As the carcasses pass an operator measurement position, an operator 290 applies an ultrasonic transducer probe from ultrasound system 218 to a specified portion of the carcass 294. Images acquired from the ultrasound system 218 are provided via connection 219 to a data acquisition system for data analysis.
In exemplary embodiments, the present invention may be used to first screen the acquired images for sufficient image quality. Next, image processing algorithms may be applied to automatically determine the fat layer boundaries, and then determine the rib locations (if visible on the image) and the top and bottom edges of the intercostales muscles. In an extension of the present invention, the fat and loin depth measurements may be used to locate one or more regions of interest of an image frame for further analysis, and selects and applies one or more image processing techniques in sequence or in parallel to the determined ROI for prediction of other muscle tissue attributes such as IMF.
Video frames are continuously captured, and processing of the captured images is implemented in response to the sensors on the transducer fixture indicating that a correct carcass skin to transducer lens face pressure range has been achieved. The pressure can be continuously monitored. Each frame for which a corresponding pressure measurement meets the pressure range criteria is evaluated for ultrasound penetration level through the first designated amount of skin (e.g., 0.69 mm for pork) as determined by histogram thresholding along the length of the probe lens. Segments of the frame at the designated depth that exceed a set reflection intensity level (e.g., 179 pixel grey scale) are gated, and regions below these segments can be excluded from the development of texture parameters. Segments of the frame at the designated depth that exceed a set reflection intensity level (e.g., 200 pixel grey scale) are gated, and any region below these segments can be excluded from a determination of subcutaneous fat depth and muscle depth. Blurred frames as detected by a wavelet transformation algorithm may be excluded from further processing of tissue texture, but may be used for subcutaneous fat depth and muscle depth.
The automation algorithm includes three subsections, each determining one of the above-mentioned boundary positions. Ultrasound image size (number of pixels along rows and columns) can vary depending on ultrasound scanner and frame grabber used for image capturing, and so the algorithm may be independent of image pixel size. The fat depth and muscle depth estimates are adjusted for the differences in ultrasound velocity in fat and muscle, respectively.
Ultrasound calibration software algorithms may be used to set image capturing parameters to a given reference. Calibration works in combination with an ultrasound scanning device, the analog video signal from the scanner, and an image frame grabber. Calibration software may be used to automatically determine if the source of the video comes from any ultrasound equipment types used in livestock scanning. Based on analysis of grey scale bars present in the images from these machines, calibration determines actual signal voltage level and compares with a 1 volt reference.
In exemplary embodiments, the present invention may be implemented as an online fat and loin depth measurement, for example, usable by a packing plant to sort pork carcasses for processing, product marketing, and paying pork producers for their harvested pigs. Systems and methods of the present invention may be employed on hot pork or beef carcasses (hot, meaning within 45 minutes postmortem), and where skin, fat, and muscle boundary determination is desired to be performed real-time so that the data can be interfaced directly with other carcass data and before the carcass leaves the hot carcass processing part in the harvesting plant.
In systems and methods of the present invention, an operator (human, automated, or combination) positions the ultrasonic probe on the skin of the carcass, and the remaining processes follow automatically, including the capture of carcass identification and live video image frames. In exemplary pork loin processing embodiments, the operator positions and maintains the ultrasound transducer (probe) fixture so that the probe is vertically aligned with and parallel to the spine or midline of the carcass, between 2 and 7 cm lateral to the midline, and on either side of the carcass. In typical packing plant environments, the carcass is vertically suspended on a trolley system. The top portion of the transducer face may be positioned so that the ultrasound image will include the last 3 to 4 ribs of the carcass.
FIG. 3 is an ultrasound image of a swine carcass loineye muscle, captured using an Aloka SSD 500V ultrasound scanner, a 12.5 cm linear transducer of 3.5 MHz and a Sensoray 2255S frame grabber. It is a longitudinal image of a swine carcass positioned over the last 3 to 4 ribs. The top-most light-grey band is the transducer skin boundary 1. Below this is a very thin light grey line which is the skin-fat boundary 2. There are further light-grey bands that correspond to three fat layers and fat-muscle layer boundary 3. The last three ribs, ribs 6, 7, and 8, respectively, are clearly seen in the lower half of the image as three vertical columns with the intercostales muscles 9 holding the ribs. The muscle above these ribs is the longissimus dorsi muscle. The boundary between the loin eye muscle and the ribs is the rib-muscle boundary 5.
A process for determining the fat depth 4 and loin eye muscle depth 10 may be automated for swine carcass data in a real time live-streaming scanning system. The fat depth 4 is the difference between the two boundary positions, skin-fat 2 and fat-muscle 3; whereas, the loin eye muscle depth 10 is the difference between the two boundary positions, fat-muscle boundary 3 and rib-muscle boundary 5. Exemplary automation algorithms for fat and loin depth are discussed in detail in the following discussions. The percentage of fat-free lean in pork muscle tissue is calculated using the fat depth and loin eye muscle depth as also discussed below.
Fat depth automation algorithms in accordance with certain embodiments include determining the two boundary positions, skin-fat and fat-muscle, from the ultrasound image of a swine carcass. FIG. 4A shows a block diagram of an example fat depth automation algorithm, which includes determining the fat-skin boundary, determining the fat-muscle boundary, and calculating the fat depth.
FIG. 5 shows a block diagram of an example skin-fat boundary automation algorithm, consistent with an example embodiment of the present invention. Threshold-based operations are used on the captured ultrasound image based on the horizontal resolution of grey level intensity to determine the desired boundary positions. From a two-dimensional (2D) grey scale image, a processing circuit calculates pixel intensity sum for each row and normalize it with a maximum sum. For instance, the sum of grey level intensity along each row (horizontal resolution) and the entire image width (typically 640 pixels) is calculated. The sum is normalized with respect to the maximum of sum value. The processing circuit then determines the row with maximum intensity sum and identifies it as the transducer-skin boundary. The intensity sum can be scanned after a set number of pixel rows (e.g., 10) from the transducer-skin boundary until the end of the rows for the skin-fat boundary. The next step involves identification of the first row after the transducer-skin interface having intensity sum over a threshold (e.g., 0.6) with a change in slope. This row value is set to be the skin-fat boundary.
FIG. 6 shows a block diagram of an example fat-muscle boundary automation algorithm, consistent with an example embodiment of the present invention. A 2D grey scale image (e.g., from of ultrasound imaging of the desired tissue area) is input to a processing circuit. The processing circuit performs multiple processing steps on different portions of the image. FIG. 6 shows one example embodiment where the processing circuit performs the processing steps on the complete image as well as left and right (subset) portions of the image. This division into three different sections is but one example. The processing circuit could perform the steps on more or less image sections by partitioning the image in a different manner. For each partition, the processing circuit calculates histogram mean for horizontal image stripes of a certain depth (e.g., 2.5 mm height) from skin-fat interface and image width that is the same as that of the ultrasound tissue area. Three peaks are identified using slope change and predefined thresholds. A comparison is performed for the determined peaks of different partitions (e.g., left, right and complete image partitions). The desired peak is then determined based on the slope of fat layers. This peak value can then be used as the fat-muscle boundary.
FIG. 7 shows a block diagram of an example intercostales muscles boundary automation algorithm, consistent with an example embodiment of the present invention. A processing circuit uses a 2D grey scale image of the ultrasound tissue area. The portion of the image used in this processing can be defined as the region vertically downwards from a determined fat-muscle boundary (e.g., from the automation algorithm of FIG. 6) and using a width that is the same as that of the ultrasound tissue area. The processing circuit calculates an intensity mean for vertical image stripes of a certain width and depth (e.g., 2.5 mm wide and from 5 mm deep) in the subregion selected. Columns are determined for all the ribs from the left side with a minimum intensity mean and change in slope. The processing circuit determines the rib top boundary rows for the rib columns by finding the local maxima with change in slope for the row intensity mean in the region 4 mm wide on either side of the rib column. Fine tuning of the average of the rib top boundaries is implemented for a pair of ribs by comparing the intensity mean difference between consecutive horizontal image stripes between the two rib columns. The strip determined is used to define the top or bottom of the intercostales muscles boundary.
An image intensity histogram mean may be computed for sliding image strips of a predefined height (e.g., 13 pixels) and width that is the same as the actual tissue area (e.g., 500 pixels), for example, moving across the rows from the skin-fat boundary to bottom with a step size equal to half the strip height (e.g., 6 pixels). The starting row of each sliding image strip and its corresponding histogram mean are stored in an array. The strips corresponding to approximately 30 mm region (e.g., strips 1 to 25) covering the upper half of an image are processed further and the strip having a local maximum greater than a specific threshold (e.g., 0.8), and with a change in slope, is determined. As such, the selected strip should have the highest histogram mean greater than the threshold in this region, and this value should be higher than its consecutive previous and next strips. All the possible strips (1/2/3) corresponding to the three fat layers, satisfying the predefined threshold and change of slope criteria, are determined and combined in a group. The starting row of the last strip in this group corresponding to the third fat layer is assigned as the average row position for the fat-muscle boundary position. Fine adjustments are performed on this boundary position to get the closest fat-muscle boundary in the region between different pairs of ribs, at the same location as that of the loin depth measurements.
An example algorithm for loin depth measurement proceeds as illustrated in the block diagram in FIG. 4B. First, the rib column positions for the first three ribs (labeled 6, 7, and 8 in FIG. 3) starting from the left side of the image are determined. Secondly, the rib top boundaries corresponding to these rib columns are calculated. Then, these rib top boundaries are processed for fine adjustment to determine the boundary of the intercostales muscles. Finally, the loin eye muscle depth is calculated using the difference between the fat-muscle and the rib-muscle boundaries and proper pixel to mm conversion ratio for the particular study setup. The depth value is adjusted for a correction factor for ultrasound velocity in muscle tissue. An accuracy flag may be assigned to each depth measurement based on the image characteristics encountered in the algorithm to get the confidence level for the measurement. Each of these steps is discussed in detail below.
Accordingly, one embodiment of the present invention relates to determining the skin-fat boundary, the fat-muscle boundary, and the intercostales muscles interface. The method involves capturing an image data from a response-provoking signal, such as an ultrasound signal. The captured image is analyzed by applying (user-defined) depth settings for preferred interface locations for fat and loin depth measurements. For example, the analysis can include the steps of a) selecting one pair of ribs for depth measurement, b) selecting multiple pairs of ribs and averaging the depth across the selected multiple ribs, and c) selecting the top or bottom interface of the intercostales muscles for loin depth.
A filtering function is applied to identify the dark vertical regions in the captured image, which can be caused by very high acoustic impedance at the skin. This can be particularly useful for compensating for dark regions that are the result of excessive scalding or burner singeing. These dark regions are then restricted from further image processing. The top interface for fat depth is the skin-fat boundary and is determined by comparing the normalized intensity sum (computed for each row across all the columns) with a predefined threshold and a change in slope criteria. The interfaces corresponding to the fat layers in the captured images are determined by comparing the grey level intensity histogram mean of selected moving horizontal image stripes with a predefined threshold. The interface for the bottommost fat layer is defined as the fat-muscle boundary. This represents the bottom interface for fat depth and the top interface for loin depth measurement.
The rib shadow column locations are identified by comparing the intensity average values calculated for vertical slices moving across the width of the image below the fat-muscle boundary. The rib top boundaries are determined by comparing the intensity average values for small horizontal slices across the each identified rib column. The bottom interface is used to determine loin depth. Depending upon (user-defined) settings, this interface can be the either top interface or the bottom interface of the intercostales muscles. This interface is determined by computing the intensity average differences for a pair of consecutive horizontal stripes starting from the average rib top boundary up to the bottom-most row of the image and between the two rib columns. The pair of strips with the lowest negative difference is identified. The first strip in this pair of strips is defined as the top interface of the intercostales muscles. The pair of strips with the highest positive difference is identified. The first strip in this pair is defined as the bottom interface of the intercostales muscles.
The results of these steps can be used for calculating the fat depth and loin depth. For example, the fat depth can be computed as a distance between the skin-fat boundary and fat-muscle boundary. The loin depth can be computed as a distance between the fat-muscle boundary and intercostales muscles boundary (e.g., top or bottom interface of intercostales muscles). A pixel scaling factor can then be applied (e.g., pixel to mm or pixel to inches) for the given system setup to determine the actual fat depth and loin depth. Moreover, a velocity correction factor can also be applied to the fat depth or loin depth to compensate for the effect of different ultrasound velocities in fat and muscle tissue.
The fat and muscle tissue of swine carcass indicated in an ultrasound image takes up only a portion of the image area (e.g., from rows 49 to 448 and columns 53 to 502 in a 640×480 pixel image). In a given image, a sub-image may be selected and considered for determining rib column positions for all the ribs from the left side of the image. Small sliding vertical strips (e.g., 10 pixels wide) are selected in the sub-image. The grey level intensity average is computed for each sliding strip. The starting column of each sliding strip and its corresponding intensity average is stored in an array. The array length is equal to the number of sliding strips in the sub-image.
The computed intensity average for sliding strips across columns is used to determine the rib column positions for the ribs starting from the left side of the image. The main focus to measure the loin depth is between a pair of ribs due to the preferable position of image ROI for texture analysis for the prediction of IMF in the same region. There are some exceptions to this where the image may be dark in this region (e.g., due to burning of the skin).
For every non-zero rib column 1 or 2, a sub-image is selected defined by the row starting from the fat-muscle boundary plus a set number of pixels, such as 120, to a set final row, such as row 420 (in mm conversion, fat-muscle boundary plus 30 mm to 105 mm). Within this sub-image a small moving image box of a set number of pixels (e.g., 13) high is selected starting from the bottom-most row. The width of this box is a set number of pixels (e.g., 30) covering the area in the vicinity of the respective rib column. The grey level intensity average of this image box is calculated. The image box is moved upwards along the rows with a step size of a set number of pixels e.g., 6) and the intensity average is computed for all the image boxes in this sub-image. The starting row of each image box and its corresponding intensity average values are stored in an array. The box having local maxima of the intensity average with a change in slope is determined for the respective rib column. The starting row of this selected box is assigned to the rib top boundary position for the respective rib. If the desired box is not found, the rib top boundary position is assigned to the starting row of the local maxima irrespective of change in slope criteria. This procedure is performed for all non-zero rib column positions to determine respective rib top boundary positions.
i) If (Rib1≠10 and Rib2≠0), then the column range is from (Rib1−step) to (Rib2+step).
ii) If (Rib1≠0 and Rib2=0), then the column range is from (Rib1−step) to (Rib 1+step).
Once the column range is decided, the row range for fine adjustment is selected to the region with row position starting from average rib top boundary minus a set number of pixels (e.g., 35) to average rib position plus a set number of pixels (e.g., 30) which is around 8 mm up and down from the average rib top boundary. Then, starting from the top row, a small image strip (e.g., 8 pixels height and width equal to the column range) is considered and its average grey level intensity is computed. The strip is moved down (e.g., using a 4 pixel step size) until the bottom row is reached, and the same computation is performed for all the strips. The starting row of each image strip and its corresponding intensity average values are stored in an array. The difference between the intensity average values for each strip with its next consecutive strip is calculated. The starting row of the strip with the lowest negative difference is assigned to the final rib-muscle boundary row position required for the loin eye muscle depth measurement. If the desired strip is not found, the final rib-muscle boundary is assigned to the average rib-top boundary. This boundary corresponds to the top interface of the intercostales muscles.
To determine the bottom interface of the intercostales muscles, the row range is selected as the region with row position starting from the rib-muscle boundary plus a set number of pixels (e.g., 24) to the rib-muscle boundary plus a set number of pixels (e.g., 70) which is approximately 18 mm down from the rib-muscle boundary. The column range is the same as the one used for fine adjustment of the rib-muscle boundary. Then, starting from the top row, a small image strip (e.g., height of 13 pixels and width equal to the column range), is considered, and its average grey level intensity is computed. The strip is moved down (e.g. using a 6 pixels step size) until the bottom row is reached, and the same computation is performed for all the strips. The starting row of each image strip and its corresponding intensity average values are stored in an array. The strip having local maxima of the intensity average with a change in slope is determined. The starting row of this selected strip is assigned to the bottom interface of the intercostales muscles. If the desired strip is not found, this boundary position is assigned to the starting row of the local maxima irrespective of the change in slope criteria. The user has the flexibility to measure the loin depth at a preferred location with respect to the intercostales muscles and the ribs. For example, one can measure loin depth up to the rib-muscle boundary (top interface of the intercostales muscles) or to the bottom interface of the intercostales muscles between any of the rib pairs.
For fine adjustment of the fat-muscle boundary, the row range for fine adjustment is selected as the region with row position starting from the average fat-muscle boundary minus a set number of pixels (e.g., 24) to the average fat boundary plus a set number of pixels (e.g., 24). This is around 6 mm up and down from the average fat-muscle boundary. The column range is the same as the one used for fine adjustment of the rib-muscle boundary. Then, starting from the top row, a small image strip (e.g., 13 pixels height and width equal to the column range), is considered, and its average grey level histogram mean is computed. The strip is moved down (e.g., using a 6 pixel step size) until the bottom row is reached, and the same computation is performed for all the strips. The starting row of each image strip and its corresponding histogram mean values are stored in an array. The difference in histogram mean values for each strip with its next consecutive strip is calculated. The starting row of the strip with the highest positive difference is assigned to the final fat-muscle boundary row position required for the fat depth measurement. If the desired strip is not found, the final fat-muscle boundary position is assigned to the average fat-muscle boundary. Once the required rib-muscle and fat-muscle boundary positions are determined, the next step calculates the loin eye muscle depth based on the two boundary positions. An accuracy flag may also be assigned to indicate measurement accuracy. The loin eye muscle depth is the difference between the two boundaries corresponding to fat-muscle (determined in fat depth automation algorithm) and rib-muscle from the previous step. This difference is divided by the pixel to mm conversion ratio (e.g., 1 mm to 3.94 pixels) for the particular setup. For example, the final loin depth formula is: Loin eye muscle depth=((Fat-muscle boundary row−rib-muscle boundary row)/3.94)*1.025974.
Table 1 presents the results of automation of backfat depth and loin eye muscle thickness measurements based on scanning 504 live animals. The images were collected using the Sensoray frame grabber and two ultrasound scanners—Aloka SSD 500V and Aquila Vet. Auto depth measurements were compared with the manual measurements done by a certified and experienced technician. The accuracy statistic for fat and loin depths is defined as an absolute difference of less than 3 mm between auto and manual measurements.
and loin depth for live pigs.

Subsets and Splits

No community queries yet

The top public SQL queries from the community will appear here once available.