Beyond pixel Superpixel-based MRI Segmentation through Traditional Machine Learning and Graph Convolutional Network

Fri, 01 Nov 2024 00:00:00 +0000

Background and Objective:

Tendon segmentation is crucial for studying tendon-related pathologies like tendinopathy, tendinosis, etc. This step further enables detailed analysis of specific tendon regions using automated or semi-automated methods. This study specifically aims at the segmentation of Achilles tendon, the largest tendon in the human body.

Methods:

This study proposes a comprehensive end-to-end tendon segmentation module composed of a preliminary superpixel-based coarse segmentation preceding the final segmentation task. The final segmentation results are obtained through two distinct approaches. In the first approach, the coarsely generated superpixels are subjected to classification using Random Forest (RF) and Support Vector Machine (SVM) classifiers to classify whether each superpixel belongs to a tendon class or not (resulting in tendon segmentation). In the second approach, the arrangements of superpixels are converted to graphs instead of being treated as conventional image grids. This classification process uses a graph-based convolutional network (GCN) to determine whether each superpixel corresponds to a tendon class or not.

Results:

All experiments are conducted on a custom-made ankle MRI dataset. The dataset comprises 76 subjects and is divided into two sets: one for training (Dataset 1, trained and evaluated using leave-one-groupout cross-validation) and the other as unseen test data (Dataset 2). Using our first approach, the final test AUC (Area Under the ROC Curve) scores using RF and SVM classifiers on the test data (Dataset 2) are 0.992 and 0.987, respectively, with sensitivities of 0.904 and 0.966. On the other hand, using our second approach (GCN-based node classification), the AUC score for the test set is 0.933 with a sensitivity of 0.899.

Conclusions:

Our proposed pipeline demonstrates the efficacy of employing superpixel generation as a coarse segmentation technique for the final tendon segmentation. Whether utilizing RF, SVM-based superpixel classification, or GCN-based classification for tendon segmentation, our system consistently achieves commendable AUC scores, especially the non-graph-based approach. Given the limited dataset, our graph-based method did not perform as well as non-graph-based superpixel classifications; however, the results obtained provide valuable insights into how well the models can distinguish between tendons and non-tendons. This opens up opportunities for further exploration and improvement.

The Role of Muscle and Tendon in Predicting Cartilage Degeneration and Tendinopathy

Fri, 28 Oct 2022 00:00:00 +0000

This study is part of an EU-funded project called P4-FIT, whose aim is innovation in tendon repair. The main objective of this study was to understand the interplay between knee cartilage, quadriceps muscle, quadriceps tendon, and patellar tendon. The dataset of another European Union project called RESTORE was used to access the CT and MRI scans of patients with knee cartilage degeneration. To gain a better understanding of the interaction between cartilage, quadriceps, and patellar tendons, several sets of features were extracted in five different categories, namely Gray-level co-occurrence matrix, Amount of fat and water containing tissues present in quadriceps and patellar tendons, Tendon thickness, Profile line analysis and Radiodensity (HU). For feature extraction, quadriceps and patellar tendons were used as the regions of interest (ROIs). Using different sets and combination of these features, different classifiers were trained to perform two distinct classification tasks. The first classifier determined whether there was degeneration of the knee cartilage, while the second one determined whether the patellar tendons were tendinopathic. To predict cartilage degeneration, some of the most important features were age, the total number of patellar tendon-containing pixels, the number of quadriceps and patellar tendon pixels containing water, etc. Using these features, our best classifier model achieved an accuracy of 89.4% for cartilage degeneration prediction. Whereas the fat-containing pixels of quadriceps and patellar tendons were two of the significant features in predicting patellar tendon involvement in tendinopathic processes. Only the sets of important features were used to obtain the best result for predicting patellar tendinopathy, which resulted in an accuracy of 83%. To our best knowledge, this is the first study to show that even without using any information about the knee bone and cartilage themselves, quadriceps and patellar tendons alone may play a powerful role in predicting knee cartilage degeneration and patellar tendinopathy. Throughout our investigation, we also found that the total amount of water and fatty tissue in the quadriceps and patellar tendons plays an important role in predicting such outcomes.

Nicola Maffulli | Zakia Khatun

Beyond pixel Superpixel-based MRI Segmentation through Traditional Machine Learning and Graph Convolutional Network

The Role of Muscle and Tendon in Predicting Cartilage Degeneration and Tendinopathy