DistilIQA: Distilling Vision Transformers for no-reference perceptual CT image quality assessment

No-reference image quality assessment (IQA) is a critical step in medical image analysis, with the objective of predicting perceptual image quality without the need for a pristine reference image. The application of no-reference IQA to CT scans is valuable in providing an automated and objective approach to assessing scan quality, optimizing radiation dose, and improving overall healthcare efficiency. In this paper, we introduce DistilIQA, a novel distilled Vision Transformer network designed for no-reference CT image quality assessment. DistilIQA integrates convolutional operations and multi-head self-attention mechanisms by incorporating a powerful convolutional stem at the beginning of the traditional ViT network. Additionally, we present a two-step distillation methodology aimed at improving network performance and efficiency. In the initial step, a "teacher ensemble network" is constructed by training five vision Transformer networks using a five-fold division schema. In the second step, a "student network", comprising of a single Vision Transformer, is trained using the original labeled dataset and the predictions generated by the teacher network as new labels. DistilIQA is evaluated in the task of quality score prediction from low-dose chest CT scans obtained from the LDCT and Projection data of the Cancer Imaging Archive, along with low-dose abdominal CT images from the LDCTIQAC2023 Grand Challenge. Our results demonstrate DistilIQA's remarkable performance in both benchmarks, surpassing the capabilities of various CNNs and Transformer architectures. Moreover, our comprehensive experimental analysis demonstrates the effectiveness of incorporating convolutional operations within the ViT architecture and highlights the advantages of our distillation methodology.

Comparative study of the methodologies used for subjective medical image quality assessment

Healthcare professionals have been increasingly viewing medical images and videos in their routine clinical practice, and this in a wide variety of environments. Both the perception and interpretation of medical visual information, across all branches of practice or medical specialties (e.g. diagnostic, therapeutic, or surgical medicine), career stages, and practice settings (e.g. emergency care), appear to be critical for patient care. However, medical images and videos are not self-explanatory and, therefore, need to be interpreted by humans, i.e. medical experts. In addition, various types of degradations and artifacts may appear during image acquisition or processing, and consequently affect medical imaging data. Such distortions tend to impact viewers' quality of experience, as well as their clinical practice. It is accordingly essential to better understand how medical experts perceive the quality of visual content. Thankfully, progress has been made in the recent literature towards such understanding. In this article, we present an up-to-date state-of the-art of relatively recent (i.e. not older than ten years old) existing studies on the subjective quality assessment of medical images and videos, as well as research works using task-based approaches. Furthermore, we discuss the merits and drawbacks of the methodologies used, and we provide recommendations about experimental designs and statistical processes to evaluate the perception of medical images and videos for future studies, which could then be used to optimise the visual experience of image readers in real clinical practice. Finally, we tackle the issue of the lack of available annotated medical image and video quality databases, which appear to be indispensable for the development of new dedicated objective metrics.