Ctrl-GenAug: Controllable Generative Augmentation for Medical Sequence Classification

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhou, Xinrui, Huang, Yuhao, Dou, Haoran, Chen, Shijing, Chang, Ao, Liu, Jia, Long, Weiran, Zheng, Jian, Xu, Erjiao, Ren, Jie, Frangi, Alejandro F., Huang, Ruobing, Cheng, Jun, Li, Xiaomeng, Xue, Wufeng, Ni, Dong
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911452469657600
author Zhou, Xinrui
Huang, Yuhao
Dou, Haoran
Chen, Shijing
Chang, Ao
Liu, Jia
Long, Weiran
Zheng, Jian
Xu, Erjiao
Ren, Jie
Frangi, Alejandro F.
Huang, Ruobing
Cheng, Jun
Li, Xiaomeng
Xue, Wufeng
Ni, Dong
author_facet Zhou, Xinrui
Huang, Yuhao
Dou, Haoran
Chen, Shijing
Chang, Ao
Liu, Jia
Long, Weiran
Zheng, Jian
Xu, Erjiao
Ren, Jie
Frangi, Alejandro F.
Huang, Ruobing
Cheng, Jun
Li, Xiaomeng
Xue, Wufeng
Ni, Dong
contents In the medical field, the limited availability of large-scale datasets and labor-intensive annotation processes hinder the performance of deep models. Diffusion-based generative augmentation approaches present a promising solution to this issue, having been proven effective in advancing downstream medical recognition tasks. Nevertheless, existing works lack sufficient semantic and sequential steerability for challenging video/3D sequence generation, and neglect quality control of noisy synthesized samples, resulting in unreliable synthetic databases and severely limiting the performance of downstream tasks. In this work, we present Ctrl-GenAug, a novel and general generative augmentation framework that enables highly semantic- and sequential-customized sequence synthesis and suppresses incorrectly synthesized samples, to aid medical sequence classification. Specifically, we first design a multimodal conditions-guided sequence generator for controllably synthesizing diagnosis-promotive samples. A sequential augmentation module is integrated to enhance the temporal/stereoscopic coherence of generated samples. Then, we propose a noisy synthetic data filter to suppress unreliable cases at semantic and sequential levels. Extensive experiments on 3 medical datasets, using 11 networks trained on 3 paradigms, comprehensively analyze the effectiveness and generality of Ctrl-GenAug, particularly in underrepresented high-risk populations and out-domain conditions.
format Preprint
id arxiv_https___arxiv_org_abs_2409_17091
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Ctrl-GenAug: Controllable Generative Augmentation for Medical Sequence Classification
Zhou, Xinrui
Huang, Yuhao
Dou, Haoran
Chen, Shijing
Chang, Ao
Liu, Jia
Long, Weiran
Zheng, Jian
Xu, Erjiao
Ren, Jie
Frangi, Alejandro F.
Huang, Ruobing
Cheng, Jun
Li, Xiaomeng
Xue, Wufeng
Ni, Dong
Computer Vision and Pattern Recognition
Artificial Intelligence
Machine Learning
In the medical field, the limited availability of large-scale datasets and labor-intensive annotation processes hinder the performance of deep models. Diffusion-based generative augmentation approaches present a promising solution to this issue, having been proven effective in advancing downstream medical recognition tasks. Nevertheless, existing works lack sufficient semantic and sequential steerability for challenging video/3D sequence generation, and neglect quality control of noisy synthesized samples, resulting in unreliable synthetic databases and severely limiting the performance of downstream tasks. In this work, we present Ctrl-GenAug, a novel and general generative augmentation framework that enables highly semantic- and sequential-customized sequence synthesis and suppresses incorrectly synthesized samples, to aid medical sequence classification. Specifically, we first design a multimodal conditions-guided sequence generator for controllably synthesizing diagnosis-promotive samples. A sequential augmentation module is integrated to enhance the temporal/stereoscopic coherence of generated samples. Then, we propose a noisy synthetic data filter to suppress unreliable cases at semantic and sequential levels. Extensive experiments on 3 medical datasets, using 11 networks trained on 3 paradigms, comprehensively analyze the effectiveness and generality of Ctrl-GenAug, particularly in underrepresented high-risk populations and out-domain conditions.
title Ctrl-GenAug: Controllable Generative Augmentation for Medical Sequence Classification
topic Computer Vision and Pattern Recognition
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2409.17091