Segmentation and Classification in Digital Pathology for Glioma Research: Challenges and Deep Learning Approaches

Kurç, Tahsin; Bakas, Spyridon; Ren, Xuhua; Bagari, Aditya; Momeni, Alexandre; Huang, Yue; Zhang, Lichi; Kumar, Ashish; Thibault, Marc; Qi, Qi; Wang, Qian; Kori, Avinash; Gevaert, Olivier; Zhang, Yun-Long; Shen, Dinggang; Khened, Mahendra; Ding, Xinghao; Krishnamurthi, Ganapathy; Kalpathy–Cramer, Jayashree; Davis, James; Zhao, Tianhao; Gupta, Rajarsi; Saltz, Joel; Farahani, Keyvan

doi:10.3389/fnins.2020.00027

Cited by 74 publications

(50 citation statements)

References 84 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Advances in machine learning allow for data to be obtained from radiographic imaging (i.e. Radiomics) and digital pathology to enhance diagnosis and predict genomic biomarkers (49)(50)(51)(52)(53). These data will likely lead to the ability of predicting tumor subtype and prognosis prior to surgical resection.…”

Section: Future Directionsmentioning

confidence: 99%

Brain Tumor Biobank Development for Precision Medicine: Role of the Neurosurgeon

et al. 2021

View full text Add to dashboard Cite

Neuro-oncology biobanks are critical for the implementation of a precision medicine program. In this perspective, we review our first year experience of a brain tumor biobank with integrated next generation sequencing. From our experience, we describe the critical role of the neurosurgeon in diagnosis, research, and precision medicine efforts. In the first year of implementation of the biobank, 117 patients (Female: 62; Male: 55) had 125 brain tumor surgeries. 75% of patients had tumors biobanked, and 16% were of minority race/ethnicity. Tumors biobanked were as follows: diffuse gliomas (45%), brain metastases (29%), meningioma (21%), and other (5%). Among biobanked patients, 100% also had next generation sequencing. Eleven patients qualified for targeted therapy based on identification of actionable gene mutations. One patient with a hereditary cancer predisposition syndrome was also identified. An iterative quality improvement process was implemented to streamline the workflow between the operating room, pathology, and the research laboratory. Dedicated tumor bank personnel in the department of neurosurgery greatly improved standard operating procedure. Intraoperative selection and processing of tumor tissue by the neurosurgeon was integral to increasing success with cell culture assays. Currently, our institutional protocol integrates standard histopathological diagnosis, next generation sequencing, and functional assays on surgical specimens to develop precision medicine protocols for our patients. This perspective reviews the critical role of neurosurgeons in brain tumor biobank implementation and success as well as future directions for enhancing precision medicine efforts.

show abstract

Section: Future Directionsmentioning

confidence: 99%

Brain Tumor Biobank Development for Precision Medicine: Role of the Neurosurgeon

et al. 2021

View full text Add to dashboard Cite

show abstract

“…The possibility of the dataset without annotation allows our algorithm to learn from the full files of slides that are presented to clinicians from real-world clinical practice, representing the full wealth of biological and technical variablitiy [ 34 ]. To the best of our knowledge, only robust studies were conducted with the deep learning approach using the publicly available digital WSI dataset in The Cancer Genome Atlas (TCGA) or The Cancer Imaging Archive (TCIA) to automate the classification of grade II, III glioma versus grade IV glioblastoma, which demonstrated up to 96% accuracy [ 35 , 36 , 37 ] However, the datasets used in the above studies are composed of many cases diagnosed before the application of the WHO’s new 2016 classification; therefore, the algorithm that was developed using the public database might not be suitable for the current WHO classification system. A recent study tried deep learning approaches for subtype classification according to the 2016 WHO classification and survival prediction using multimodal magnetic resonance images of a brain tumor [ 38 ].…”

Section: Discussionmentioning

confidence: 99%

Classification of Diffuse Glioma Subtype from Clinical-Grade Pathological Images Using Deep Transfer Learning

Hyeon

Rha

et al. 2021

Sensors

View full text Add to dashboard Cite

Diffuse gliomas are the most common primary brain tumors and they vary considerably in their morphology, location, genetic alterations, and response to therapy. In 2016, the World Health Organization (WHO) provided new guidelines for making an integrated diagnosis that incorporates both morphologic and molecular features to diffuse gliomas. In this study, we demonstrate how deep learning approaches can be used for an automatic classification of glioma subtypes and grading using whole-slide images that were obtained from routine clinical practice. A deep transfer learning method using the ResNet50V2 model was trained to classify subtypes and grades of diffuse gliomas according to the WHO’s new 2016 classification. The balanced accuracy of the diffuse glioma subtype classification model with majority voting was 0.8727. These results highlight an emerging role of deep learning in the future practice of pathologic diagnosis.

show abstract

“…Our last approach, which uses deep learning, applies a mask region-based convolutional neural network (R-CNN). Many digital pathology approaches have used mask R-CNNs in applications of nuclei segmentation 12 and cancer [35][36][37] . Similar to our study, these studies spatially identified specific features (e.g., nuclei).…”

Section: Discussionmentioning

confidence: 99%

Spectroscopic and Deep Learning-Based Approaches to Identify and Quantify Cerebral Microhemorrhages

Crouzet

Jeong

Chae

et al. 2021

Preprint

View full text Add to dashboard Cite

Cerebral microhemorrhages (CMHs) are associated with cerebrovascular disease, cognitive impairment, and normal aging. One method to study CMHs is to analyze histological sections (5-40 μm) stained with Prussian blue. Currently, users manually and subjectively identify and quantify Prussian blue-stained regions of interest, which is prone to inter-individual variability and can lead to significant delays in data analysis. To improve this labor-intensive process, we developed and compared three digital pathology approaches to identify and quantify CMHs from Prussian blue-stained brain sections: 1) ratiometric analysis of RGB pixel values, 2) phasor analysis of RGB images, and 3) deep learning using a mask region-based convolutional neural network. We applied these approaches to a preclinical mouse model of inflammation-induced CMHs. One-hundred CMHs were imaged using a 20x objective and RGB color camera. To determine the ground truth, four users independently annotated Prussian blue-labeled CMHs. The deep learning and ratiometric approaches performed better than the phasor analysis approach compared to the ground truth. The deep learning approach had the most precision of the three methods. The ratiometric approach has the most versatility and maintained accuracy, albeit with less precision. Our data suggest that implementing these methods to analyze CMH images can drastically increase the processing speed while maintaining precision and accuracy.

show abstract

Segmentation and Classification in Digital Pathology for Glioma Research: Challenges and Deep Learning Approaches

Cited by 74 publications

References 84 publications

Brain Tumor Biobank Development for Precision Medicine: Role of the Neurosurgeon

Brain Tumor Biobank Development for Precision Medicine: Role of the Neurosurgeon

Classification of Diffuse Glioma Subtype from Clinical-Grade Pathological Images Using Deep Transfer Learning

Spectroscopic and Deep Learning-Based Approaches to Identify and Quantify Cerebral Microhemorrhages

Contact Info

Product

Resources

About