Interpretable Deep Learning Identified the Significance of 1 Gy Volume on Lymphopenia after Radiotherapy in Breast Cancer

医学 乳腺癌 放射治疗 剂量体积直方图 核医学 临床意义 癌症 肿瘤科 放射科 放射治疗计划 内科学
作者
Fang Chen,Ping Zhou,K.W. Lee,Qing Liu,Aya El Helali,Jian‐Yue Jin,A.W.M. Lee,Hao Yu,Feng‐Ming Kong
出处
期刊:International Journal of Radiation Oncology Biology Physics [Elsevier]
卷期号:117 (2): e168-e168
标识
DOI:10.1016/j.ijrobp.2023.06.1006
摘要

Lymphopenia is common after radiotherapy (RT) and is known for its significance on poor survival outcomes in patients with breast cancer. Previous work has demonstrated the significance of point dosimetric factors like the volume receiving 5 Gy. Considering the full dosimetric data together, this study aimed to develop and validate predictive models for lymphopenia after RT in breast cancer.Patients with breast cancer treated with radiation therapy in adjuvant setting and with complete dosimetric data were eligible. Combining dose-volume histogram (DVH) dosimetric and clinical factors, dense neural network (DNN) models were developed to predict both the reduction in lymphocyte counts and the graded lymphopenia in breast cancer patients after adjuvant RT. A Shapley additive explanation was applied to explain each feature's directional contributions. The generalization of DNN models was validated in both internal and independent external validation cohorts. P<0.05 was considered to be significant.A total of 928 consecutive patients with invasive breast cancer were eligible for this study. Treatment volumes of nearly all irradiation dose levels of DVH were significant predictors for lymphopenia after RT, including volumes at very low-dose 1 Gy (V1) of all structures considered including the lung, heart and body. DNN models using full DVH dosimetric and clinical factors were built and a simplified model was further established and validated in both internal and external validation cohorts. This simplified DNN AI model, combining full DVH dosimetric parameters of all OARs and five key clinical factors including baseline lymphocyte counts, tumor stage, RT technique, RT fields and RT fractionation, showed a predictive accuracy of 77% and above.This study demonstrated and externally validated the significance of an AI model of combining clinical and full dosimetric data, especially the volume of low dose at as low as 1 Gy of all critical structures on lymphopenia after RT in patients with breast cancer. The significance of V1 deserves special attention, as modern arc RT technology often has relatively high value of this parameter. Further study warranted for breast cancer RT plan optimization.
最长约 10秒,即可获得该文献文件

科研通智能强力驱动
Strongly Powered by AbleSci AI
科研通是完全免费的文献互助平台,具备全网最快的应助速度,最高的求助完成率。 对每一个文献求助,科研通都将尽心尽力,给求助人一个满意的交代。
实时播报
没有神的过往完成签到,获得积分10
1秒前
1秒前
2秒前
3秒前
3秒前
芋圆不圆完成签到,获得积分10
4秒前
招财不肥发布了新的文献求助10
5秒前
zxc111发布了新的文献求助10
5秒前
魔幻的从梦完成签到,获得积分10
5秒前
6秒前
Xiaoxiao应助sunyexuan采纳,获得10
7秒前
8秒前
9秒前
淼淼之锋完成签到 ,获得积分10
9秒前
赢赢完成签到 ,获得积分10
9秒前
10秒前
11秒前
科目三应助落落采纳,获得10
13秒前
67发布了新的文献求助10
13秒前
13秒前
溜溜完成签到,获得积分10
13秒前
xixi完成签到 ,获得积分10
14秒前
wanci应助科研通管家采纳,获得10
14秒前
撒上咖啡应助科研通管家采纳,获得10
14秒前
RC_Wang应助科研通管家采纳,获得10
14秒前
JamesPei应助科研通管家采纳,获得10
14秒前
酷波er应助科研通管家采纳,获得10
14秒前
琪琪扬扬发布了新的文献求助10
14秒前
sutharsons应助科研通管家采纳,获得30
14秒前
orixero应助科研通管家采纳,获得10
15秒前
研友_VZG7GZ应助科研通管家采纳,获得10
15秒前
科研通AI5应助科研通管家采纳,获得10
15秒前
清爽老九应助科研通管家采纳,获得20
15秒前
酷波er应助科研通管家采纳,获得10
15秒前
wanci应助科研通管家采纳,获得10
15秒前
香蕉觅云应助科研通管家采纳,获得10
15秒前
赘婿应助科研通管家采纳,获得10
15秒前
hui发布了新的文献求助30
15秒前
传奇3应助科研通管家采纳,获得10
15秒前
15秒前
高分求助中
Continuum Thermodynamics and Material Modelling 3000
Production Logging: Theoretical and Interpretive Elements 2700
Social media impact on athlete mental health: #RealityCheck 1020
Ensartinib (Ensacove) for Non-Small Cell Lung Cancer 1000
Unseen Mendieta: The Unpublished Works of Ana Mendieta 1000
Bacterial collagenases and their clinical applications 800
El viaje de una vida: Memorias de María Lecea 800
热门求助领域 (近24小时)
化学 材料科学 生物 医学 工程类 有机化学 生物化学 物理 纳米技术 计算机科学 内科学 化学工程 复合材料 基因 遗传学 物理化学 催化作用 量子力学 光电子学 冶金
热门帖子
关注 科研通微信公众号,转发送积分 3527928
求助须知:如何正确求助?哪些是违规求助? 3108040
关于积分的说明 9287614
捐赠科研通 2805836
什么是DOI,文献DOI怎么找? 1540070
邀请新用户注册赠送积分活动 716904
科研通“疑难数据库(出版商)”最低求助积分说明 709808