The role of fractional-order calculus and continuous-time random-walk diffusion model in the differentiation of benign and malignant head and neck lesions
10.3969/j.issn.1002-1671.2025.02.007
- VernacularTitle:分数阶微积分和连续时间随机游走扩散模型在鉴别头颈部良恶性病变的应用
- Author:
Jun LIU
1
;
Yi'nan SUN
;
Li HUA
;
Qing YANG
;
Fei WANG
;
Hualin YANG
;
Ming CHEN
;
Qiuyang GUO
;
Mengxiao LIU
;
Juan ZHU
Author Information
1. 安庆市立医院(安徽医科大学安庆医学中心)医学影像科,安徽 安庆 246003
- Publication Type:Journal Article
- Keywords:
head and neck lesions;
fractional-order calculus;
continuous-time random-walk;
readout-segmented
- From:
Journal of Practical Radiology
2025;41(2):206-210
- CountryChina
- Language:Chinese
-
Abstract:
Objective To investigate the value of fractional-order calculus(FROC)and continuous-time random-walk(CTRW)diffusion models based on readout segmentation of long variable echo-trains(RESOLVE)in identifying benign and malignant lesions in the head and neck.Methods A retrospective analysis was conducted on 61 patients pathologically confirmed head and neck lesions,including 19 benign lesions(BL)and 42 malignant lesions(ML).The ML were further divided into a lymphoma subgroup(LS)with 9 cases(14 lesions)and a non-lymphoma malignant lesion subgroup(MLS)with 33 cases.The parameters of DFROC,βFROC,μFROC,DCTRW,αCTRW and βCTRW were obtained from the two diffusion models;Independent sample t-tests or U tests were used to compare the differences in each parameter between benign and malignant groups and among various subgroups,and the receiver operating characteristic(ROC)curve was used to analyze the diagnostic efficacy of each parameter.Area under the curve(AUC)was compared by DeLong test.Results DFROC,μFROC,DCTRW and αCTRW showed significant differences between benign and malignant,BL and LS,BL and MLS and LS and MLS,with αCTRW showed the highest diagnostic efficacy;βFROC showed differences between BL and LS,BL and MLS,whileβCTRW did not show differences between benign and malignant groups,and among subgroups.Conclusion FROC and CTRW diffusion models based on RESOLVE can distinguish between benign and malignant head and neck lesions with multiple parameters,and provide metrics reflecting tissue heterogeneity.
