{"title":"Lite-FBCN:从核磁共振成像图像进行脑疾病分类的轻量级快速双线性卷积网络","authors":"Dewinda Julianensi Rumala, Reza Fuad Rachmadi, Anggraini Dwi Sensusiati, I Ketut Eddy Purnama","doi":"arxiv-2409.10952","DOIUrl":null,"url":null,"abstract":"Achieving high accuracy with computational efficiency in brain disease\nclassification from Magnetic Resonance Imaging (MRI) scans is challenging,\nparticularly when both coarse and fine-grained distinctions are crucial.\nCurrent deep learning methods often struggle to balance accuracy with\ncomputational demands. We propose Lite-FBCN, a novel Lightweight Fast Bilinear\nConvolutional Network designed to address this issue. Unlike traditional\ndual-network bilinear models, Lite-FBCN utilizes a single-network architecture,\nsignificantly reducing computational load. Lite-FBCN leverages lightweight,\npre-trained CNNs fine-tuned to extract relevant features and incorporates a\nchannel reducer layer before bilinear pooling, minimizing feature map\ndimensionality and resulting in a compact bilinear vector. Extensive\nevaluations on cross-validation and hold-out data demonstrate that Lite-FBCN\nnot only surpasses baseline CNNs but also outperforms existing bilinear models.\nLite-FBCN with MobileNetV1 attains 98.10% accuracy in cross-validation and\n69.37% on hold-out data (a 3% improvement over the baseline). UMAP\nvisualizations further confirm its effectiveness in distinguishing closely\nrelated brain disease classes. Moreover, its optimal trade-off between\nperformance and computational efficiency positions Lite-FBCN as a promising\nsolution for enhancing diagnostic capabilities in resource-constrained and or\nreal-time clinical environments.","PeriodicalId":501289,"journal":{"name":"arXiv - EE - Image and Video Processing","volume":null,"pages":null},"PeriodicalIF":0.0000,"publicationDate":"2024-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Lite-FBCN: Lightweight Fast Bilinear Convolutional Network for Brain Disease Classification from MRI Image\",\"authors\":\"Dewinda Julianensi Rumala, Reza Fuad Rachmadi, Anggraini Dwi Sensusiati, I Ketut Eddy Purnama\",\"doi\":\"arxiv-2409.10952\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Achieving high accuracy with computational efficiency in brain disease\\nclassification from Magnetic Resonance Imaging (MRI) scans is challenging,\\nparticularly when both coarse and fine-grained distinctions are crucial.\\nCurrent deep learning methods often struggle to balance accuracy with\\ncomputational demands. We propose Lite-FBCN, a novel Lightweight Fast Bilinear\\nConvolutional Network designed to address this issue. Unlike traditional\\ndual-network bilinear models, Lite-FBCN utilizes a single-network architecture,\\nsignificantly reducing computational load. Lite-FBCN leverages lightweight,\\npre-trained CNNs fine-tuned to extract relevant features and incorporates a\\nchannel reducer layer before bilinear pooling, minimizing feature map\\ndimensionality and resulting in a compact bilinear vector. Extensive\\nevaluations on cross-validation and hold-out data demonstrate that Lite-FBCN\\nnot only surpasses baseline CNNs but also outperforms existing bilinear models.\\nLite-FBCN with MobileNetV1 attains 98.10% accuracy in cross-validation and\\n69.37% on hold-out data (a 3% improvement over the baseline). UMAP\\nvisualizations further confirm its effectiveness in distinguishing closely\\nrelated brain disease classes. Moreover, its optimal trade-off between\\nperformance and computational efficiency positions Lite-FBCN as a promising\\nsolution for enhancing diagnostic capabilities in resource-constrained and or\\nreal-time clinical environments.\",\"PeriodicalId\":501289,\"journal\":{\"name\":\"arXiv - EE - Image and Video Processing\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2024-09-17\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"arXiv - EE - Image and Video Processing\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/arxiv-2409.10952\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"arXiv - EE - Image and Video Processing","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/arxiv-2409.10952","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
Lite-FBCN: Lightweight Fast Bilinear Convolutional Network for Brain Disease Classification from MRI Image
Achieving high accuracy with computational efficiency in brain disease
classification from Magnetic Resonance Imaging (MRI) scans is challenging,
particularly when both coarse and fine-grained distinctions are crucial.
Current deep learning methods often struggle to balance accuracy with
computational demands. We propose Lite-FBCN, a novel Lightweight Fast Bilinear
Convolutional Network designed to address this issue. Unlike traditional
dual-network bilinear models, Lite-FBCN utilizes a single-network architecture,
significantly reducing computational load. Lite-FBCN leverages lightweight,
pre-trained CNNs fine-tuned to extract relevant features and incorporates a
channel reducer layer before bilinear pooling, minimizing feature map
dimensionality and resulting in a compact bilinear vector. Extensive
evaluations on cross-validation and hold-out data demonstrate that Lite-FBCN
not only surpasses baseline CNNs but also outperforms existing bilinear models.
Lite-FBCN with MobileNetV1 attains 98.10% accuracy in cross-validation and
69.37% on hold-out data (a 3% improvement over the baseline). UMAP
visualizations further confirm its effectiveness in distinguishing closely
related brain disease classes. Moreover, its optimal trade-off between
performance and computational efficiency positions Lite-FBCN as a promising
solution for enhancing diagnostic capabilities in resource-constrained and or
real-time clinical environments.