Diffuse large B‐cell lymphoma segmentation in PET‐CT images via hybrid learning for feature fusion

Abstract

Purpose: Diffuse large B-cell lymphoma (DLBCL) is an aggressive type of lymphoma with high mortality and poor prognosis that especially has a high incidence in Asia. Accurate segmentation of DLBCL lesions is crucial for clinical radiation therapy. However, manual delineation of DLBCL lesions is tedious and time-consuming. Automatic segmentation provides an alternative solution but is difficult for diffuse lesions without the sufficient utilization of multimodality information. Our work is the first study focusing on positron emission tomography and computed tomography (PET-CT) feature fusion for the DLBCL segmentation issue. We aim to improve the fusion performance of complementary information contained in PET-CT imaging with a hybrid learning module in the supervised convolutional neural network. Methods: First, two encoder branches extract single-modality features, respectively. Next, the hybrid learning component utilizes them to generate spatial fusion maps which can quantify the contribution of complementary information. Such feature fusion maps are then concatenated with specific-modality (i.e., PET and CT) feature maps to obtain a representation of the final fused feature maps in different scales. Finally, the reconstruction part of our network creates a prediction map of DLBCL lesions by integrating and up-sampling the final fused feature maps from encoder blocks in different scales. Results: The ability of our method was evaluated to detect foreground and segment lesions in three independent body regions (nasopharynx, chest, and abdomen) of a set of 45 PET-CT scans. Extensive ablation experiments compared our method to four baseline techniques for multimodality fusion (input-level (IL) fusion, multichannel (MC) strategy, multibranch (MB) strategy, and quantitative weighting (QW) fusion). The results showed that our method achieved a high detection accuracy (99.63% in the nasopharynx, 99.51% in the chest, and 99.21% in the abdomen) and had the superiority in segmentation performance with the mean dice similarity coefficient (DSC) of 73.03% and the modified Hausdorff distance (MHD) of 4.39 mm, when compared with the baselines (DSC: IL:

The application of machine learning to imaging in hematological oncology: A scoping review

“…A total of 11 studies applied ML to segmentation tasks ( Table 2 ) in FDG-PET/CT images: 3 applied it to DLBCL ( 57 , 58 , 67 ), 1 to DLBCL and HL ( 78 ), 1 to DLBCL and NHL ( 64 ), 1 to natural killer (NK)/T-cell lymphoma ( 60 ), 1 to HL ( 65 ), and 4 to NHL/lymphoma unspecified ( 59 , 61 , 63 , 66 ). One study applied ML to MRI images to segment PCNSL ( 62 ).…”

“…CNN-based methods were most commonly applied to segmentation tasks (eight studies ( 57 , 58 , 61 – 65 , 67 ),), but RFs ( 59 ), adversarial networks ( 60 ), and conditional random fields ( 66 ) were also used.…”

“…Models were validated using random splits ( 60 ), cross-validation approaches ( 59 , 64 , 65 , 67 ), separate (but not fully independent) datasets ( 57 , 61 , 63 ), or validation was unreported or not performed ( 62 , 66 ).…”

“…Each study selected a number of different ground truths for comparison with the ML results, including manual selection after 41% SUV max thresholding (57, 58), 41% SUV max thresholding in manually placed ROIs (59,64,66), manual segmentation by radiologists but without any further details of the methodology (60)(61)(62)(63)65), or not stated (67).…”

Multimodal deep learning model on interim [18F]FDG PET/CT for predicting primary treatment failure in diffuse large B-cell lymphoma

“…Forth (IV), the model learned the spatial contribution of feature maps from PET and CT encoders by a quantitative weighting strategy which calculated the convolutional result as a weighted matrix [36]. In the last model (V), PET and CT features extracted from two identical encoders were combined by the hybrid learning approach, a modal fusion method we published before [37], which generated spatial fusion maps and quanti ed the contribution of complementary information. These fusion maps were then concatenated with speci c-modality (i.e.…”