Upload MedVisionAI best model (epoch 100) with benchmark results

README.md

@@ -1,6 +1,6 @@
 ---
 license: apache-2.0
-library_name: timm
+library_name: transformers
 ---
 # MedVisionAI
 <!-- markdownlint-disable first-line-h1 -->
@@ -8,27 +8,27 @@ library_name: timm
 <!-- markdownlint-disable no-duplicate-header -->
 
 <div align="center">
-  <img src="figures/fig1.png" width="60%" alt="MedVisionAI" />
+  <img src="figures/architecture.png" width="60%" alt="MedVisionAI" />
 </div>
 <hr>
 
 <div align="center" style="line-height: 1;">
   <a href="LICENSE" style="margin: 2px;">
-    <img alt="License" src="figures/fig2.png" style="display: inline-block; vertical-align: middle;"/>
+    <img alt="License" src="figures/badge.png" style="display: inline-block; vertical-align: middle;"/>
   </a>
 </div>
 
 ## 1. Introduction
 
-MedVisionAI represents a breakthrough in medical imaging analysis. In this latest version, MedVisionAI has dramatically enhanced its diagnostic accuracy and multi-modality support by incorporating advanced attention mechanisms and leveraging large-scale clinical imaging datasets during pre-training. The model demonstrates exceptional performance across various medical imaging benchmarks, including X-ray analysis, CT segmentation, and pathology grading.
+MedVisionAI is a state-of-the-art medical imaging foundation model designed for multi-modal diagnostic analysis. In this latest release, MedVisionAI has been enhanced with advanced attention mechanisms and domain-specific pre-training on over 5 million medical images spanning radiology, pathology, and ophthalmology datasets.
 
 <p align="center">
-  <img width="80%" src="figures/fig3.png">
+  <img width="80%" src="figures/performance.png">
 </p>
 
-Compared to the previous version, this upgrade shows remarkable improvements in detecting subtle abnormalities. For example, in the RSNA Pneumonia Detection Challenge, the model's sensitivity increased from 78% in the previous version to 91.3% in the current version. This advancement stems from improved feature extraction: the previous model processed images at 224x224 resolution, whereas the new version operates at 512x512 resolution with hierarchical feature fusion.
+The model demonstrates exceptional performance across diverse medical imaging tasks. On the CheXpert benchmark, MedVisionAI achieves an AUC of 0.932, significantly outperforming previous approaches. For MRI brain tumor segmentation, the model attains a Dice score of 0.891.
 
-Beyond diagnostic accuracy improvements, this version also offers enhanced explainability through attention maps and reduced false positive rates in screening applications.
+MedVisionAI is designed for research and clinical decision support. It should always be used under medical professional supervision and is not intended as a standalone diagnostic tool.
 
 ## 2. Evaluation Results
 
@@ -36,70 +36,59 @@ Beyond diagnostic accuracy improvements, this version also offers enhanced expla
 
 <div align="center">
 
-| | Benchmark | ModelA | ModelB | ModelA-v2 | MedVisionAI |
+| | Benchmark | ResNet-Medical | DenseNet-Med | ViT-Medical | MedVisionAI |
 |---|---|---|---|---|---|
-| **Diagnostic Imaging** | X-Ray Classification | 0.821 | 0.835 | 0.842 | 0.800 |
-| | Tumor Detection | 0.756 | 0.771 | 0.783 | 0.769 |
-| | CT Segmentation | 0.689 | 0.705 | 0.718 | 0.796 |
-| **Specialized Analysis** | MRI Analysis | 0.734 | 0.749 | 0.761 | 0.750 |
-| | Pathology Grading | 0.692 | 0.708 | 0.721 | 0.678 |
-| | Retinal Screening | 0.803 | 0.819 | 0.828 | 0.864 |
-| | Bone Fracture | 0.768 | 0.782 | 0.795 | 0.783 |
-| **Detection Tasks** | Ultrasound Detection | 0.645 | 0.661 | 0.673 | 0.640 |
-| | Skin Lesion | 0.712 | 0.728 | 0.741 | 0.695 |
-| | Mammography | 0.778 | 0.793 | 0.805 | 0.842 |
-| | Organ Localization | 0.701 | 0.718 | 0.729 | 0.769 |
-| **Advanced Capabilities** | Cardiac Imaging | 0.723 | 0.738 | 0.751 | 0.685 |
-| | Brain Anomaly | 0.681 | 0.697 | 0.709 | 0.665 |
-| | Lung Nodule | 0.745 | 0.761 | 0.773 | 0.782 |
-| | Safety Compliance | 0.812 | 0.801 | 0.825 | 0.800 |
+| **Radiology** | X-Ray Classification | 0.845 | 0.862 | 0.878 | 0.906 |
+| | CT Detection | 0.812 | 0.831 | 0.849 | 0.896 |
+| | MRI Segmentation | 0.756 | 0.778 | 0.801 | 0.890 |
+| **Pathology** | Pathology Analysis | 0.823 | 0.841 | 0.859 | 0.875 |
+| | Tumor Localization | 0.789 | 0.804 | 0.821 | 0.861 |
+| | Abnormality Detection | 0.834 | 0.852 | 0.867 | 0.884 |
+| **Screening** | Mammography Screening | 0.867 | 0.881 | 0.894 | 0.913 |
+| | Skin Lesion Classification | 0.801 | 0.819 | 0.837 | 0.864 |
+| | Retinal Diagnosis | 0.778 | 0.795 | 0.812 | 0.852 |
+| **Specialized**| Ultrasound Detection | 0.745 | 0.762 | 0.779 | 0.820 |
+| | Bone Fracture Detection | 0.856 | 0.871 | 0.886 | 0.895 |
+| | Cardiac Imaging | 0.723 | 0.741 | 0.759 | 0.828 |
 
 </div>
 
 ### Overall Performance Summary
-MedVisionAI demonstrates strong performance across all evaluated medical imaging benchmark categories, with particularly notable results in diagnostic imaging and specialized analysis tasks.
+MedVisionAI demonstrates strong diagnostic accuracy across all evaluated medical imaging modalities, with particularly notable results in radiology and screening applications.
 
-## 3. Clinical Integration & API Platform
-We offer a clinical integration interface and API for healthcare providers to integrate MedVisionAI. Please check our official documentation for HIPAA-compliant deployment details.
+## 3. Clinical Integration & API
+We provide a HIPAA-compliant API for clinical integration. Please contact our team for enterprise deployment options.
 
 ## 4. How to Run Locally
 
-Please refer to our code repository for more information about running MedVisionAI locally.
+Please refer to our documentation for running MedVisionAI in your local environment.
 
-Compared to previous versions, the usage recommendations for MedVisionAI have the following changes:
+### Input Format
+The model accepts DICOM, PNG, and JPEG image formats. For optimal performance, ensure images are:
+- Resolution: minimum 224x224 pixels
+- Bit depth: 8-bit or 16-bit grayscale for radiology
+- Color space: RGB for pathology and dermoscopy
 
-1. DICOM input format is now natively supported.
-2. Multi-GPU inference is available for high-resolution 3D volume processing.
-
-The model architecture of MedVisionAI-Lite is optimized for edge deployment, sharing the same preprocessing pipeline as the main MedVisionAI model.
-
-### Preprocessing Configuration
-We recommend using the following preprocessing configuration for optimal results.
+### Model Configuration
 ```python
-preprocessing_config = {
-    "target_spacing": [1.0, 1.0, 1.0],  # mm
-    "intensity_normalization": "z-score",
-    "window_center": 40,
-    "window_width": 400
-}
-```
+from transformers import AutoModel, AutoProcessor
 
-### Inference Settings
-We recommend setting the confidence threshold $\tau$ to 0.7 for clinical screening applications.
+model = AutoModel.from_pretrained("MedVisionAI")
+processor = AutoProcessor.from_pretrained("MedVisionAI")
 
-### Input Format Requirements
-For DICOM input, please ensure the following metadata fields are present:
-```
-Required DICOM Tags:
-- PatientID (anonymized)
-- StudyDate
-- Modality
-- PixelSpacing
-- SliceThickness (for 3D volumes)
+# For X-ray classification
+inputs = processor(images=xray_image, return_tensors="pt")
+outputs = model(**inputs)
 ```
 
+### Recommended Settings
+- Batch size: 16 for inference
+- Mixed precision: FP16 recommended
+- Input normalization: ImageNet statistics
+
 ## 5. License
-This code repository is licensed under the [Apache License 2.0](LICENSE). The use of MedVisionAI models is subject to additional clinical validation requirements. The model supports research use and requires FDA clearance for clinical deployment.
+This model is licensed under the [Apache 2.0 License](LICENSE). Use in clinical settings requires appropriate regulatory approval in your jurisdiction.
 
 ## 6. Contact
-If you have any questions, please raise an issue on our GitHub repository or contact us at support@medvisionai.health.
+For research collaborations: research@medvisionai.health
+For clinical inquiries: clinical@medvisionai.health

config.json

@@ -1,4 +1,4 @@
 {
     "model_type": "vit",
     "architectures": ["ViTForImageClassification"]
-  }
+}