Extracting 2D weak labels from volume labels using multiple instance learning in CT hemorrhage detection

Abstract

Multiple instance learning (MIL) is a supervised learning methodology that aims to allow models to learn instance class labels from bag class labels, where a bag is defined to contain multiple instances. MIL is gaining traction for learning from weak labels but has not been widely applied to 3D medical imaging. MIL is well-suited to clinical CT acquisitions since (1) the highly anisotropic voxels hinder application of traditional 3D networks and (2) patch-based networks have limited ability to learn whole volume labels. In this work, we apply MIL with a deep convolutional neural network to identify whether clinical CT head image volumes possess one or more large hemorrhages (> 20cm 3 ), resulting in a learned 2D model without the need for 2D slice annotations. Individual image volumes are considered separate bags, and the slices in each volume are instances. Such a framework sets the stage for incorporating information obtained in clinical reports to help train a 2D segmentation approach. Within this context, we evaluate the data requirements to enable generalization of MIL by varying the amount of training data. Our results show that a training size of at least 400 patient image volumes was needed to achieve accurate per-slice hemorrhage detection. Over a five-fold cross-validation, the leading model, which made use of the maximum number of training volumes, had an average true positive rate of 98.10%, an average true negative rate of 99.36%, and an average precision of 0.9698. The models have been made available along with source code 1 to enabled continued exploration and adaption of MIL in CT neuroimaging.

Automatic Estimation of Ulcerative Colitis Severity from Endoscopy Videos using Ordinal Multi-Instance Learning

“…This framework has shown effective performance across image modalities like digital histopathology (Campanella et al 2019) and chest X-ray (Schwab et al 2020). Multi-class formulations have also been explored (Remedios et al 2020;Lu et al 2021).…”

Deep multiscale convolutional feature learning for intracranial hemorrhage classification and weakly supervised localization

“…Bar et al [ 17 ] improved classification results by combining the previously independent segmentation and classification tasks with dependencies with relatively few datasets. Remedios et al [ 18 ] utilized multi-instance learning to learn features from weak labels to identify massive bleeds.…”

Combining Attention-Based Multiple Instance Learning and Gaussian Processes for CT Hemorrhage Detection

“…A scan is classified as ICH if at least one slice in this scan has ICH and is normal if all slices are normal. Few studies use MIL method in ICH detection [15,16]. For instance, Remedios et al [15] combine CNNs with MIL to predict ICH at scan-level, but the model was trained with a max-pooling operation so it could only select the most positive instance in a bag.…”