Computer-assisted Applications in Medicine (CAiM) Group at ETH Zurich is led by Prof. Dr. Orcun Goksel. CAiM works on computer-based processing and interpretation of medical images, focusing on the following topics:
To learn more about the individual topics and to show a list of related projects, please select a topic below:
Current projects in Computer-assisted Applications in Medicine include:
Bone and joint disorders are a leading cause of physical disability worldwide and account for 50 % of all chronical diseases in people over 50 years of age. The goal of this subproject is to produce patient-specific models, in particular of musculoskeletal system.
With a focus on orthopedic surgeries, these models will allow for simulating pathological conditions and functional movement capabilities with the immediate intention of pre-operative planning and future potential use in diagnosis.
University Hospital Balgrist
Bone and joint disorders are a leading cause of physical disability worldwide and account for 50 % of all chronical diseases in people over 50 years of age. The goal of this project is to devise computerized simulation techniques that will assist in the preoperative planning of othopaedic surgeries.
The shoulder and the forearm are our current research focus.
Ultrasound is a low-cost, non-ionizing, interactive imaging modality, used commonly in clinics as an invaluable tool for screening, diagnostics, and therapy guidance. However, its image formation physics and interactive nature makes it substantially user-dependent, for instance some pathology diagnosed by one sonographer may be missed by another one, due to poor transducer navigation or image interpretation. Therefore, ultrasound examinations require a high level of expertise, involving several technical and practical skills, which usually requires years to reach a sufficient level of competence.
In particular in OB/GYN, in-vivo training opportunities of navigation skills with volunteers for educational and residency training programs are also limited by the discomfort involved, e.g. in transvaginal examinations. Also, rare clinical case are not always available. Current ultrasound training systems lack image realism, prohibiting the training of image interpretation and limiting the training of navigation skills.
In this project, we aim at developing virtual-reality simulators for ultrasound training. In such a life-like simulation, trainees shall steer a (mock) transducer over a (mock) torso while watching the simulated image scenery on the (ultrasound) screen and being assessed via objective (computed) metrics. We focus on OB/GYN applications due to difficulties in finding volunteers, its large market potential, and the outstanding expertise of our team in this field. Nevertheless, the acquired know-how and developed techniques will translate to other anatomical targets and clinical scenarios in the future.
Tissue elasticity has been used in medicine as an indication of tissue anomalies for centuries. Palpation of breast and prostate are still the most common clinical procedures in cancer screening for those corresponding anatomical structures. Elastography is an emerging technique for imaging tissue viscoelasticity. This project aims to address some of the pending issues in generating ultrasound elastography images, while also investigating novel acquisition and analysis techniques for advanced imaging applications.
Elastography is an emerging technique for imaging tissue viscoelasticity from internal tissue displacements observed often in magnetic resonance or ultrasound time sequences. Elastography based on ultrasound is particularly important, as ultrasound is real-time, cost-effective, non-ionizing, and thus a widely-used imaging modality. Elastography has great potential in diagnosis and treatment planning because pathologies, e.g. tumors and cirrhosis, are often stiffer than their surrounding tissue. It can furthermore be beneficial as an indication of tissue composition for robust characterization and segmentation of anatomical structures with different mechanical properties, e.g. for arterial plaque characterization. Despite all these benefits, elastography is still not fully exploited in clinics. Several different methodological approaches proposed in this field are still not fully mature and there is a clear need for robust methods for imaging tissue elasticity in an accurate, fast, and reproducible way. This project aims to address some of the pending issues in generating ultrasound elastography images, while also investigating novel acquisition and analysis techniques for advanced imaging applications.
Different imaging techniques and modalities provide additional information to physicians for better diagnosis and screening, and tissue elasticity is one such information that clinicians can exploit in their decision making. Results of our research will improve current clinical practices by helping to bring tissue elasticity imaging into the arsenal of everyday medicine.
The goal of this project is to investigate and develop unsupervised or minimally supervised techniques for the segmentation of the anatomy (in particular, bone structures).
Statistical models of variability in anatomical structures are essential for a wide range of medical applications, e.g., to reconstruct missing information in implant design or help localize bones for radiotherapy planning. Segmentation of the anatomy is a necessary step towards such robust statistical shape and intensity modeling. A major obstacle in generating such statistical models is the need for a sufficiently large amount of correctly segmented medical data, from which the statistical model can be derived. Besides the purposes of active shape modelling based segmentation, it is beneficial to develop prior-less or weakly-supervised techniques to prepare the initial data used for modelling, and to deal with cases not well described by the statistical shape model (e.g. trauma, pathology).
Conventional segmentation methods for such large sets on segmentation data require either a significant amount of manual, interactive effort of the medical personnel or a statistical shape model of the bone in question, and sometimes a combination of both. The objective of this project is to develop weakly supervised methods for segmentation for subsequent automatic generation of statistical models. To this end, we are utilizing single atlas methods to provide semantic context in a general segmentation framework. We are particularly interested in bone segmentation to be utilized in the generation of a statistical skeleton atlas of the body.