Segmentation of orthopaedic images of the knee for implant identification and characterization
Research internship, 6 months
Context and objectives
The analysis of orthopaedic X-rays images is challenging after the implantation of metallic prostheses such as Total Knee Arthroplasty (TKA) or Total Hip Arthroplasty (THA). Many different implants are available worldwide, having different designs, based on different concepts (cemented or uncemented) and using different biomaterials (titanium or chromium-cobalt alloy). These variations induce different fixation patterns with varying aspects on postoperative X-ray images (bone-implant interface, implant positioning, bone remodelling, etc.). Refer to the illustrations at the end of the attached document.
The reliability of the interpretation of orthopaedic x-ray images therefore depends largely on the specialization of Medical Doctors (MD) and on their knowledge of implants and of surgical details. Only highly specialized MDs, mainly arthroplasty surgeons, can accurately analyze postoperative X-ray images and diagnose implant failures. The medical community in general remains unable to make accurate and relevant interpretations. This generates a loss of efficiency and resources for the healthcare system and patients, especially for those living far from healthcare centers. Therefore, there is nowadays a strong need for the development of a platform for the automatic interpretation of orthopaedic images to assist in the diagnosis of articular pathologies and the analysis of prostheses.
The proposed research internship constitutes the first step towards such a platform, which is the objective of a collaborative project between the Santy Orthopaedic Clinic based in Lyon and the Creatis Medical Imaging Research Center, which is part of the Lyon University, bringing together medical and methodological expertise of involved MDs and researchers. The project will initially focus on prosthesis analysis of the knee (total, unicompartmental, revision) for the purposes of identifying the type of the implant and characterizing its status (position, wear, loosening, presence of osteophytes, radiolucencies etc.). This investigation will require a precise delimitation of the surfaces of the prosthesis and those of nearby tissues, which is usually carried out via image processing techniques, namely object segmentation.
Recent progress in machine learning methods, and particularly those of supervised deep learning methods, has considerably advanced the state-of-the-art in medical image analysis [1]. Deep convolutional networks in particular create hierarchies of feature representations from images and have been shown to be very effective in the accurate segmentation of multiple objects in medical images [2-5]. These methods can be applied to the segmentation of knee tissues and implants for subsequent measurement purposes. Santy clinicians will provide sufficiently large annotated datasets covering pathological and normal cases to train and validate models.
Required profile
We are looking for a motivated collaborator capable of critical thinking, taking initiatives and able to work autonomously as well as in a collective setting, having interest for medical imaging and good sense of responsibility (and humor). The candidate should be studying towards completing a master degree in computer science or a related engineering field. She should have a solid background in applied mathematics, image processing and computer science, in addition to good programming skills, preferably in Python programming language. A working knowledge of deep learning methods is necessary.
Application
We encourage interested candidates to send us their resume accompanied by a cover letter and a transcript of recent grades.
Salary
The intern will me remunerated at a rate of 700 to 1000 € / month depending on her profile.
References
- Litjens, G., Kooi, T., Bejnordi, B. E., Setio, A. A. A., Ciompi, F., Ghafoorian, M., ... & Sánchez, C. I. (2017). A survey on deep learning in medical image analysis. Medical image analysis, 42, 60-88.
- Hu, P., Wu, F., Peng, J., Bao, Y., Chen, F., & Kong, D. (2017). Automatic abdominal multi-organ segmentation using deep convolutional neural network and time-implicit level sets. International journal of computer assisted radiology and surgery, 12(3), 399-411.
- Roth, H. R., Shen, C., Oda, H., Oda, M., Hayashi, Y., Misawa, K., & Mori, K. (2018). Deep learning and its application to medical image segmentation. Medical Imaging Technology, 36(2), 63-71.
- Gibson, E., Giganti, F., Hu, Y., Bonmati, E., Bandula, S., Gurusamy, K., ... & Barratt, D. C. (2018). Automatic multi-organ segmentation on abdominal CT with dense v-networks. IEEE transactions on medical imaging, 37(8), 1822-1834.
- Heinrich, M. P., Oktay, O., & Bouteldja, N. (2019). OBELISK-Net: Fewer layers to solve 3D multi-organ segmentation with sparse deformable convolutions. Medical image analysis, 54, 1-9.