Abstract:BackgroundAccurate and precise detection of brain lesions on MR images (MRI) is paramount for accurately relating lesion location to impaired behavior. In this paper, we present a novel method to automatically detect brain lesions from a T1-weighted 3D MRI. The proposed method combines the advantages of both unsupervised and supervised methods.MethodsFirst, unsupervised methods perform a unified segmentation normalization to warp images from the native space into a standard space and to generate probability ma… Show more
“…Moreover, the lesions present a high degree of discontinuity which was a big challenge to this algorithm, thereby resulting in a very poor performance as can be seen in Table 1 where INNN17 has a 0.732 overlap and INNN21 has a 0.511 performance. We also compared the proposed method against the Automated Lesion Detection on MRI Scans Using Combined Unsupervised and Supervised Methods by Guo et al [34] and Multiplicative Intrinsic Component Optimization (MICO) [35] method by Li et al, and their results are reported in Figures 11, 12, and 13.…”
Section: Evaluation and Experimental Resultsmentioning
We develop a swift, robust, and practical tool for detecting brain lesions with minimal user intervention to assist clinicians and researchers in the diagnosis process, radiosurgery planning, and assessment of the patient's response to the therapy. We propose a unified gravitational fuzzy clustering-based segmentation algorithm, which integrates the Newtonian concept of gravity into fuzzy clustering. We first perform fuzzy rule-based image enhancement on our database which is comprised of T1/T2 weighted magnetic resonance (MR) and fluid-attenuated inversion recovery (FLAIR) images to facilitate a smoother segmentation. The scalar output obtained is fed into a gravitational fuzzy clustering algorithm, which separates healthy structures from the unhealthy. Finally, the lesion contour is automatically outlined through the initialization-free level set evolution method. An advantage of this lesion detection algorithm is its precision and its simultaneous use of features computed from the intensity properties of the MR scan in a cascading pattern, which makes the computation fast, robust, and self-contained. Furthermore, we validate our algorithm with large-scale experiments using clinical and synthetic brain lesion datasets. As a result, an 84%–93% overlap performance is obtained, with an emphasis on robustness with respect to different and heterogeneous types of lesion and a swift computation time.
“…Moreover, the lesions present a high degree of discontinuity which was a big challenge to this algorithm, thereby resulting in a very poor performance as can be seen in Table 1 where INNN17 has a 0.732 overlap and INNN21 has a 0.511 performance. We also compared the proposed method against the Automated Lesion Detection on MRI Scans Using Combined Unsupervised and Supervised Methods by Guo et al [34] and Multiplicative Intrinsic Component Optimization (MICO) [35] method by Li et al, and their results are reported in Figures 11, 12, and 13.…”
Section: Evaluation and Experimental Resultsmentioning
We develop a swift, robust, and practical tool for detecting brain lesions with minimal user intervention to assist clinicians and researchers in the diagnosis process, radiosurgery planning, and assessment of the patient's response to the therapy. We propose a unified gravitational fuzzy clustering-based segmentation algorithm, which integrates the Newtonian concept of gravity into fuzzy clustering. We first perform fuzzy rule-based image enhancement on our database which is comprised of T1/T2 weighted magnetic resonance (MR) and fluid-attenuated inversion recovery (FLAIR) images to facilitate a smoother segmentation. The scalar output obtained is fed into a gravitational fuzzy clustering algorithm, which separates healthy structures from the unhealthy. Finally, the lesion contour is automatically outlined through the initialization-free level set evolution method. An advantage of this lesion detection algorithm is its precision and its simultaneous use of features computed from the intensity properties of the MR scan in a cascading pattern, which makes the computation fast, robust, and self-contained. Furthermore, we validate our algorithm with large-scale experiments using clinical and synthetic brain lesion datasets. As a result, an 84%–93% overlap performance is obtained, with an emphasis on robustness with respect to different and heterogeneous types of lesion and a swift computation time.
“…43 The combined unsupervised and supervised components along with SVM classi¯er achieved an average Dice coe±cient of 73.1% for detecting stroke lesion in T1-weighted MRIs. 44 An automated approach based on unsupervised classi¯cation with fuzzy c-means clustering with the self-adjusted intensity thresholds detected cerebral infarct lesions with a DSI of 89.9%. 45 An automated approach based on Bayesian MRF was successfully used to segment stroke lesion in FLAIR MRI images with a Dice similarity coe±cient of 0.60.…”
Neuroimaging investigation is an essential parameter to detect infarct lesion in stroke patients. Precise detection of brain lesions is an important task related to impaired behavior. In this paper, we aimed to develop an automatic method to segment and classify infarct lesion in di®usion-weighted imaging (DWI) of brain MRI. The method includes hybrid fuzzy c-means (HFCM) clustering in which the structure of c-means clustering is modi¯ed with rough sets and fuzzy sets to improve the segmentation performance with selfadjusted intensity thresholds. Quantitative evaluation was carried out on 128 MRI slices of brain image collected from ischemic stroke patients at the Department of Radiology, IMS and SUM Hospital, Bhubaneswar. The informative statistical features have been extracted using gray-level co-occurrence matrix (GLCM) and used to classify the types of stroke infarct according to the Oxfordshire Community Stroke Project (OCSP) classi¯cation. The parameters such as accuracy, Dice similarity index (DSI) and Jaccard index (JI) were utilized to evaluate the e®ectiveness of the proposed method in detecting the stroke lesions. The segmentation method achieved the average accuracy, DSI and JI of 96.8%, 95.8% and 92.2%, respectively, in support vector machine (SVM) classi¯er. The obtained results are higher in terms of random forest (RF) classi¯cation. With a high Dice coe±cient of
“…The SVM is considered as one of the supervised learning models used in various applications such as segmentation, object recognition, speaker identification, and medical diagnosis (10,38). It was used to classify voxels into normal and pathological tissue (45)(46)(47).…”
Context: Medical imaging technologies are an indispensable tool in medicine today developed to satisfy the significant demand for information on medical imaging by visualizing internal organs for clinical analysis. This enables the radiologists and clinicians to accurately understand the patient's condition and makes medical practices easier, more effective for patients, and cheaper for the healthcare system. Objective: The current study aimed at presenting a comprehensive review on the recent classification and segmentation techniques of brain tumors in magnetic resonance image (MRI). Data Source: Google Scholar, ScienceDirect, Web of Knowledge, Springer, and manual search of reference lists from 1990 to 2018. Inclusion Criteria: The current study considered brain tumors since they are relatively less common and more important compared with other tumors due to their high morbidity rate. Results: Many automated brain tumors segmentation algorithms of magnetic resonance imaging (MRI) were reviewed and discussed including their advantages and limitations to provide a clear insight into these algorithms. The review concentratedon the state-of-art methods of segmentation of MRI brain tumors since they attracted a significant attention in the recent two decades resulting in many algorithms being developed for automated, semi-automated, and interactive segmentation of brain tumors. While there is a significant development of segmentation algorithms, they are rarely used clinically due to lack of interaction between developers and clinicians. Conclusions: Most studies did not consider grading of brain tumors and did not distinguish to which grade the brain tumor belonged. This enables the developers to understand how the margins of brain tumors appear in medical images. Limitations: The most important limitations that make brain tumors segmentation remaina challenging task are the variety of the shape and intensity of tumors in addition to the probability of inhomogeneity of tumorous tissue.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.