Evaluation of Enhanced Learning Techniques for Segmenting Ischaemic Stroke Lesions in Brain Magnetic Resonance Perfusion Images Using a Convolutional Neural Network Scheme

Malla, Carlos Uziel Perez; Hernández, María del C. Valdés; Rachmadi, Muhammad Febrian; Komura, Taku

doi:10.3389/fninf.2019.00033

Cited by 22 publications

(5 citation statements)

References 45 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Some papers have attempted to directly segment the perfusion images using complex neural networks in ischemic stroke lesions in perfusion MRI and intracranial arteries and veins in 4D cerebral CT perfusion. Both papers reported reasonable success which may be attributable to the blood or contrast highlighting nature of perfusion images (Mendrik et al 2010, Malla et al 2019. With respect to deep learning, the challenging aspect of utilizing perfusion images is the accurate registration of MRI contours to the perfusion image as the trained model will largely be affected by the quality of the input data.…”

Section: Segmentation Taskmentioning

confidence: 99%

Brain tumor segmentation using 3D Mask R-CNN for dynamic susceptibility contrast enhanced perfusion imaging

Jeong,

Lei,

Kahn

et al. 2020

Phys. Med. Biol.

View full text Add to dashboard Cite

The segmentation of neoplasms is an important part of radiotherapy treatment planning, monitoring disease progression, and predicting patient outcome. In the brain, functional magnetic resonance imaging (MRI) like dynamic susceptibility contrast enhanced (DSCE) or T1-weighted dynamic contrast enhanced (DCE) perfusion MRI are important tools for diagnosis. They play a crucial role in providing pre-operative assessment of tumor histology, grading, and biopsy guidance. However, the manual contouring of these neoplasms is tedious, expensive, time-consuming, and vulnerable to inter-observer variability. In this work, we propose a 3D mask region-based convolutional neural network (R-CNN) method to automatically segment brain tumors in DSCE MRI perfusion images. As our goal is to simultaneously localize and segment the tumor, our training process contained both a region-of-interest (ROI) localization and regression with voxel-wise segmentation. The combination of classification loss, ROI location and size regression loss, and segmentation loss were used to supervise the proposed network. We retrospectively investigated 21 patients’ perfusion images, with between 50 and 70 perfusion time point volumes, a total of 1260 3D volumes. Tumor contours were automatically segmented by our proposed method and compared against other state-of-the-art methods and those delineated by physicians as the ground truth. The results of our method demonstrated good agreement with the ground truth contours. The average DSC, precision, recall, Hausdorff distance, mean surface distance (MSD), root MSD, and center of mass distance were 0.90 ± 0.04, 0.91 ± 0.04, 0.90 ± 0.06, 7.16 ± 5.78 mm, 0.45 ± 0.34 mm, 1.03 ± 0.72 mm, and 0.86 ± 0.91 mm, respectively. These results support the feasibility of our method in accurately localizing and segmenting brain tumors in DSCE perfusion MRI. Our 3D Mask R-CNN segmentation method in DSCE perfusion imaging has great promise for future clinical use.

show abstract

Section: Segmentation Taskmentioning

confidence: 99%

Brain tumor segmentation using 3D Mask R-CNN for dynamic susceptibility contrast enhanced perfusion imaging

Jeong,

Lei,

Kahn

et al. 2020

Phys. Med. Biol.

View full text Add to dashboard Cite

show abstract

“…In some studies, CT segmentation applications of the stroke-related region are carried out with different deep-learning models [13,17,[41][42][43][44]. There are also studies that use MR images, which are acquired over a longer duration compared to CT images, to perform stroke classification [45][46][47][48][49][50][51][52]. The number of studies classifying normal, ischemia, and hemorrhage CT images is limited [9,12,14,23].…”

Section: Resultsmentioning

confidence: 99%

Performance Evaluation of Different Deep Learning Models for Classifying Ischemic, Hemorrhagic, and Normal Computed Tomography Images: Transfer Learning Approaches

Altıntaş,

Öziç

2024

KONJES

View full text Add to dashboard Cite

A stroke is a case of damage to a brain area due to a sudden decrease or complete cessation of blood flow to the brain. The interruption or reduction of the transportation of oxygen and nutrients through the bloodstream causes damage to brain tissues. Thus, motor or sensory impairments occur in the body part controlled by the affected area of the brain. There are primarily two main types of strokes: ischemic and hemorrhagic. When a patient is suspected of having a stroke, a computed tomography scan is performed to identify any tissue damage and facilitate prompt intervention quickly. Early intervention can prevent the patient from being permanently disabled throughout their lifetime. This study classified ischemic, hemorrhage, and normal computed tomography images taken from international databases as open source with AlexNet, ResNet50, GoogleNet, InceptionV3, ShuffleNet, and SqueezeNet deep learning models using transfer learning approach. The data were divided into 80% training and 20% testing, and evaluation metrics were calculated by five-fold cross-validation. The best performance results for the three-class output were obtained with AlexNet as 0.9086±0.02 precision, 0.9097±0.02 sensitivity, 0.9091±0.02 F1 score, 0.9089±0.02 accuracy. The average area under curve values was obtained with AlexNet 0.9920±0.005 for ischemia, 0.9828±0.008 for hemorrhage, and 0.9686±0.012 for normal.

show abstract

“…For example, Figure 2 shows 19 studies in the PubMed database retrieved for the terms, "neuroimaging", "stroke", and "convolutional neural network". Excluding review papers, a study actually treating Alzheimer's patients and a study that did not use neuroimaging data for modeling, the majority of retrieved studies used CNNs for detecting stroke, segmenting lesions, white matter hyperintensities, or other regions of the brain that may be affected by atrophy [28][29][30][31][32][33][34][35][36][37][38][39] (N = 12; see Figure S7 and Table S3). Two studies relied on CNNs to accelerate sequence acquisition or increase sequence resolution 40,41 , and another two studies leveraged CNNs for outcome prediction, but in the acute setting 22,42 .…”

Section: Recent Work Has Indicated Cnns Can Adequately Discriminate T...mentioning

confidence: 99%

Distinct brain morphometry patterns revealed by deep learning improve prediction of aphasia severity

Teghipco

Newman‐Norlund

Fridriksson

et al. 2023

Preprint

View full text Add to dashboard Cite

Emerging evidence suggests that post-stroke aphasia severity depends on the integrity of the brain beyond the stroke lesion. While measures of lesion anatomy and brain integrity combine synergistically to explain aphasic symptoms, significant interindividual variability remains unaccounted for. A possible explanatory factor may be the spatial distribution of brain atrophy beyond the lesion. This includes not just the specific brain areas showing atrophy, but also distinct three-dimensional patterns of atrophy. Here, we tested whether deep learning with Convolutional Neural Networks (CNN) on whole brain morphometry (i.e., segmented tissue volumes) and lesion anatomy can better predict which individuals with chronic stroke (N=231) have severe aphasia, and whether encoding spatial dependencies in the data might be capable of improving predictions by identifying unique individualized spatial patterns. We observed that CNN achieves significantly higher accuracy and F1 scores than Support Vector Machine (SVM), even when the SVM is nonlinear or integrates linear and nonlinear dimensionality reduction techniques. Performance parity was only achieved when the SVM was directly trained on the latent features learned by the CNN. Saliency maps demonstrated that the CNN leveraged widely distributed patterns of brain atrophy predictive of aphasia severity, whereas the SVM focused almost exclusively on the area around the lesion. Ensemble clustering of CNN saliency maps revealed distinct morphometry patterns that were unrelated to lesion size, highly consistent across individuals, and implicated unique brain networks associated with different cognitive processes as measured by the wider neuroimaging literature. Individualized predictions of severity depended on both ipsilateral and contralateral features outside of the location of stroke. Our findings illustrate that three-dimensional network distributions of atrophy in individuals with aphasia are directly associated with aphasia severity, underscoring the potential for deep learning to improve prognostication of behavioral outcomes from neuroimaging data, and highlighting the prospective benefits of interrogating spatial dependence at different scales in multivariate feature space.

show abstract

Evaluation of Enhanced Learning Techniques for Segmenting Ischaemic Stroke Lesions in Brain Magnetic Resonance Perfusion Images Using a Convolutional Neural Network Scheme

Cited by 22 publications

References 45 publications

Brain tumor segmentation using 3D Mask R-CNN for dynamic susceptibility contrast enhanced perfusion imaging

Brain tumor segmentation using 3D Mask R-CNN for dynamic susceptibility contrast enhanced perfusion imaging

Performance Evaluation of Different Deep Learning Models for Classifying Ischemic, Hemorrhagic, and Normal Computed Tomography Images: Transfer Learning Approaches

Distinct brain morphometry patterns revealed by deep learning improve prediction of aphasia severity

Contact Info

Product

Resources

About