Construction and evaluation of an automatic scoring and grading visualization system for inter-slice contour irregularity in target volumes

Purpose To mitigate inter-slice contour irregularity in lattice radiotherapy (LRT) and replace subjective inspection, we developed an automated tool that flags candidate slices for physician review. Methods DICOM-RT files were processed in Python to extract target contours. Inter-slice irregularity was quantified by CD, ACR, DSC, and HD95. After min-max normalization and weighting, a composite score (S) was calculated. Result In the illustrative LRT case (left scapular tumor), slice 207 had a composite irregularity score S of 0.5285 and was classified as extremely high irregularity; this adjacent-slice transition showed a decrease in the Dice coefficient to 0.9145, an increase in HD95 to 25.39 mm, and a 13.57% expansion in cross-sectional area, while S rapidly returned to 0.0394 at slice 208, indicating that the abnormality was localized and non-propagating. This slice appeared as an isolated peak in the trend graph, suggesting that it may reflect delineation inconsistency or local anatomical variation and warrants further physician review. Across the entire sequence, low-irregularity slices accounted for 74.55%, and only 1.21% were classified as extremely high irregularity. After contour refinement, the GTV volume increased by 4.58%, the number of accommodated lattice vertices increased from 18 to 20, V1200 improved from 94.99% to 96.40%, and PVDR remained essentially unchanged (2.23 vs. 2.24). In the five independent test cases, the tool flagged a total of 25 slices with S ≥ 0.3; consensus physician review identified 22 slices requiring correction, of which 19 were covered by the tool (coverage rate 86.4%), and 19 of the 25 flagged slices were confirmed as requiring correction (confirmation rate 76.0%). After contour correction, the mean S decreased from 0.4038 to 0.3345. Based on the distribution of irregularity grades in the 35-case training set, a preliminary irregularity tolerance framework was proposed for high-precision, mobile, and low-precision target volumes. Conclusion The automated tool enables objective, accurate, and reproducible evaluation of inter-slice irregularity, effectively identifying and flagging critical slices for physician review. Physician validation confirmed that correcting flagged slices significantly improves contour consistency, demonstrating the tool’s potential to enhance the reliability and efficiency of geometric quality assurance in radiotherapy planning.

Authors

Institutions

Publication Details

Journal
Frontiers in Oncology
Published
2026-11-17
DOI
https://doi.org/10.3389/fonc.2026.1933589
Primary Topic
Advanced Radiotherapy Techniques
Type
article
Field-Weighted Citation Impact
0.00

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Construction and evaluation of an automatic scoring and grading visualization system for inter-slice contour irregularity in target volumes

Chuan Chen, Peng Zhou, Mengyuan Si, Linshan Li
3 links
Frontiers in Oncology
Advanced Radiotherapy Techniques
article

Construction and evaluation of an automatic scoring and grading visualization system for inter-slice contour irregularity in target volumes

Chuan Chen, Peng Zhou, Mengyuan Si, Linshan Li
article en

Abstract

Purpose To mitigate inter-slice contour irregularity in lattice radiotherapy (LRT) and replace subjective inspection, we developed an automated tool that flags candidate slices for physician review. Methods DICOM-RT files were processed in Python to extract target contours. Inter-slice irregularity was quantified by CD, ACR, DSC, and HD95. After min-max normalization and weighting, a composite score (S) was calculated. Result In the illustrative LRT case (left scapular tumor), slice 207 had a composite irregularity score S of 0.5285 and was classified as extremely high irregularity; this adjacent-slice transition showed a decrease in the Dice coefficient to 0.9145, an increase in HD95 to 25.39 mm, and a 13.57% expansion in cross-sectional area, while S rapidly returned to 0.0394 at slice 208, indicating that the abnormality was localized and non-propagating. This slice appeared as an isolated peak in the trend graph, suggesting that it may reflect delineation inconsistency or local anatomical variation and warrants further physician review. Across the entire sequence, low-irregularity slices accounted for 74.55%, and only 1.21% were classified as extremely high irregularity. After contour refinement, the GTV volume increased by 4.58%, the number of accommodated lattice vertices increased from 18 to 20, V1200 improved from 94.99% to 96.40%, and PVDR remained essentially unchanged (2.23 vs. 2.24). In the five independent test cases, the tool flagged a total of 25 slices with S ≥ 0.3; consensus physician review identified 22 slices requiring correction, of which 19 were covered by the tool (coverage rate 86.4%), and 19 of the 25 flagged slices were confirmed as requiring correction (confirmation rate 76.0%). After contour correction, the mean S decreased from 0.4038 to 0.3345. Based on the distribution of irregularity grades in the 35-case training set, a preliminary irregularity tolerance framework was proposed for high-precision, mobile, and low-precision target volumes. Conclusion The automated tool enables objective, accurate, and reproducible evaluation of inter-slice irregularity, effectively identifying and flagging critical slices for physician review. Physician validation confirmed that correcting flagged slices significantly improves contour consistency, demonstrating the tool’s potential to enhance the reliability and efficiency of geometric quality assurance in radiotherapy planning.

Frontiers in OncologyVol. 16
Army Medical University (CN), Daping Hospital (CN)
Army Medical University, Natural Science Foundation Project of Chongqing, Chongqing Science and Technology Commission
Good health and well-being
Advanced Radiotherapy Techniques
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon