Adaptive IoU Thresholding for Improving Small Object Detection: A Proof-of-Concept Study of Hand Erosions Classification of Patients with Rheumatic Arthritis on X-ray Images

Abstract

In recent years, much research evaluating the radiographic destruction of finger joints in patients with rheumatoid arthritis (RA) using deep learning models was conducted. Unfortunately, most previous models were not clinically applicable due to the small object regions as well as the close spatial relationship. In recent years, a new network structure called RetinaNets, in combination with the focal loss function, proved reliable for detecting even small objects. Therefore, the study aimed to increase the recognition performance to a clinically valuable level by proposing an innovative approach with adaptive changes in intersection over union (IoU) values during training of Retina Networks using the focal loss error function. To this end, the erosion score was determined using the Sharp van der Heijde (SvH) metric on 300 conventional radiographs from 119 patients with RA. Subsequently, a standard RetinaNet with different IoU values as well as adaptively modified IoU values were trained and compared in terms of accuracy, mean average accuracy (mAP), and IoU. With the proposed approach of adaptive IoU values during training, erosion detection accuracy could be improved to 94% and an mAP of 0.81 ± 0.18. In contrast Retina networks with static IoU values achieved only an accuracy of 80% and an mAP of 0.43 ± 0.24. Thus, adaptive adjustment of IoU values during training is a simple and effective method to increase the recognition accuracy of small objects such as finger and wrist joints.

Automated Radiographic Total Sharp Score (ARTSS) in Rheumatoid Arthritis: A Solution to Reduce Inter-Intra Reader Variation and Enhancing Clinical Practice

“…By utilizing the ResNet50 architecture for orientation and a combination of U-Net and YOLOv7 for segmentation and joint identification, we achieved an accuracy of 99%. This surpasses the 87% accuracy achieved by Radke KL, et al [25] using RetinaNet, and aligns closely with Chaturvedi N, who reported a near-perfect identification rate using RetinaNet, as presented in Table 2.…”

Deep Learning Model Based on You Only Look Once Algorithm for Detection and Visualization of Fracture Areas in Three-Dimensional Skeletal Images

“…The term “Area of Intersection” refers to the region corresponding to the correct answer, while “Area of Union” pertains to the predicted area. The IoU can achieve a maximum value of 1, with a higher value signifying a more precise prediction [ 27 ].…”

A Stepwise Approach to Analyzing Musculoskeletal Imaging Data With Artificial Intelligence

“…Although developed AI models can save time when completing tasks, developing models takes a significant amount of time. 25 For example, Mulford et al created a pelvic annotation model that is able to label 21 structures on an anteroposterior pelvis image. In order to create this model, it required 1,100 images to be annotated with the 21 structures.…”

“…The model places a bounding box around the specific joint with erosion, numbers the specific joint, and places a specific erosion value on the joint. 25 Image segmentation goes a step further than object detection and places a "mask" over the object of interest. Although object detection places a bounding box around the object, that box contains both the object as well as background information.…”

“…For example, Radke et al developed an object detection model capable of detecting joint erosion in a hand radiograph. The model places a bounding box around the specific joint with erosion, numbers the specific joint, and places a specific erosion value on the joint 25 …”