File size: 4,577 Bytes
1a80759 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 | import gradio as gr
import cv2
import numpy as np
import math
from ultralytics import YOLO
# Load Model
MODEL_PATH = "best.pt"
model = YOLO(MODEL_PATH, task="pose")
def get_hip_points(image_crop, model, conf=0.25):
results = model(image_crop, conf=conf, verbose=False)
for r in results:
if r.keypoints and r.keypoints.shape[1] >= 2:
kpts = r.keypoints.xy[0].cpu().numpy().astype(int)
valid_kpts = [p for p in kpts if p[0] != 0 and p[1] != 0]
if len(valid_kpts) >= 2:
valid_kpts.sort(key=lambda p: p[1])
return valid_kpts[-1], valid_kpts[0]
return None, None
def calculate_slope_angle(center, rim, other_center):
try:
if other_center[0] - center[0] == 0: m_h = 0
else: m_h = (other_center[1] - center[1]) / (other_center[0] - center[0])
if rim[0] - center[0] == 0: m_r = 1e9
else: m_r = (rim[1] - center[1]) / (rim[0] - center[0])
tan_theta = abs((m_r - m_h) / (1 + m_h * m_r))
return math.degrees(math.atan(tan_theta))
except: return 0.0
def get_diagnosis_style(angle):
if angle >= 30: return "Dysplasia (High Severity)", "#FF4B4B"
if angle >= 25: return "Borderline / Mild", "#FFA500"
return "Normal", "#09AB3B"
def analyze_xray(input_img, conf):
image = np.array(input_img)
image = cv2.cvtColor(image, cv2.COLOR_RGB2BGR)
h, w, _ = image.shape
mid_x = w // 2
right_hip_img = image[:, :mid_x]
left_hip_img = image[:, mid_x:]
r_tri, r_rim = get_hip_points(right_hip_img, model, conf)
l_tri_local, l_rim_local = get_hip_points(left_hip_img, model, conf)
if r_tri is None or l_tri_local is None:
return image, "⚠️ **Error:** Could not detect anatomical landmarks. Try adjusting the confidence threshold."
l_tri_global = (l_tri_local[0] + mid_x, l_tri_local[1])
l_rim_global = (l_rim_local[0] + mid_x, l_rim_local[1])
# Draw Hilgenreiner's line (Base Line)
cv2.line(image, tuple(r_tri), tuple(l_tri_global), (0, 255, 255), 3)
ang_r = calculate_slope_angle(r_tri, r_rim, l_tri_global)
diag_r, color_r = get_diagnosis_style(ang_r)
ang_l = calculate_slope_angle(l_tri_global, l_rim_global, r_tri)
diag_l, color_l = get_diagnosis_style(ang_l)
# Formatting the output text with Markdown
result_text = f"### 🩺 Clinical Assessment Results\n\n"
result_text += f"**Right Hip AI:** `{ang_r:.1f}°` ➔ **{diag_r}**\n\n"
result_text += f"**Left Hip AI:** `{ang_l:.1f}°` ➔ **{diag_l}**"
# Draw Roof Lines
cv2.line(image, tuple(r_tri), tuple(r_rim), (0, 0, 255), 3)
cv2.line(image, tuple(l_tri_global), tuple(l_rim_global), (0, 0, 255), 3)
return cv2.cvtColor(image, cv2.COLOR_BGR2RGB), result_text
# ==========================================
# Gradio UI Design (Enhanced & English)
# ==========================================
custom_theme = gr.themes.Soft(
primary_hue="blue",
secondary_hue="slate",
)
with gr.Blocks(title="DislocPred: DDH System", theme=custom_theme) as demo:
# Header Section
with gr.Row():
gr.Markdown(
"""
# 🏥 DislocPred: AI Diagnostic Assistant
**Automated Detection & Grading for Developmental Dysplasia of the Hip (DDH)** Upload a pediatric pelvic X-ray (AP view). The system will analyze the anatomical landmarks, calculate the Acetabular Index (AI), and provide a clinical evaluation.
"""
)
# Main Application Area
with gr.Row():
# Left Column: Inputs
with gr.Column(scale=1):
gr.Markdown("### 📥 Input Panel")
input_img = gr.Image(type="pil", label="Upload X-ray Image")
with gr.Accordion("⚙️ Advanced Model Settings", open=False):
conf_slider = gr.Slider(0.1, 1.0, value=0.25, step=0.05, label="Detection Confidence Threshold")
btn = gr.Button("🔍 Run AI Analysis", variant="primary")
# Right Column: Outputs
with gr.Column(scale=1):
gr.Markdown("### 📤 Diagnostic Output")
output_img = gr.Image(label="Annotated Radiograph")
output_text = gr.Markdown(label="Results")
# Connect logic to UI
btn.click(
fn=analyze_xray,
inputs=[input_img, conf_slider],
outputs=[output_img, output_text]
)
# Launch with proper Hugging Face settings
if __name__ == "__main__":
demo.launch(server_name="0.0.0.0", server_port=7860) |