University of Twente Student Theses


Recognition of crystals in synovial fluids by a neural network.

Willemsen, B. (2021) Recognition of crystals in synovial fluids by a neural network.

Full text not available from this repository.

Full Text Status:Access to this publication is restricted
Abstract:Crystal classification with a polarised light microscope is the gold standard for making diagnosis of several types of inflammatory arthritis. Rheumatologists extract synovial fluid from patients and this fluid is examined. The visible crystals are classified and this classification determines the patient's treatment. However, rheumatologists often fail to properly classify crystals, which can lead to suboptimal treatment of their patients. This report shows that using convolutional neural networks is a promising method to classify crystals. The neural network reached a sensitivity of 89.7\% for Gout and 88.6\% for Calcium Pyrophosphate Dihydrate on a data set comprised of images from Gout, Tophageous Gout, Calcium Pyrophosphate Dihydrate, Triamcinolon, artifacts and healthy cells. This shows that a non-optimized network was able to reach a performance similar to professionals who currently perform this classification in routine care. If the current drawbacks are further analysed, it is plausible that a fully optimized network will be able to assist professionals in achieving more accurate diagnosis.
Item Type:Essay (Bachelor)
Faculty:TNW: Science and Technology
Subject:44 medicine, 54 computer science
Programme:Biomedical Technology BSc (56226)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page