TY - JOUR

T1 - Multi-Class Surface Generation of Complex Anatomical Structures Using Neural Networks

AU - Epifanov, R. U.I.

AU - Федотова, Яна Валерьевна

AU - Popov, D. R.

AU - Мулляджанов, Рустам Илхамович

N1 - Glazkova A. V., Smal I., Lyashevskaya O., Morozov D. Contextual normalization of abbreviations with large language models // Doklady Mathematics. — 2026. — Vol. 112. - № 1. — P. 227–233. — DOI: 10.1134/S1064562425700231. This work was supported by the Mathematical Center in Akademgorodok, agreement no. 075-15-2025-349 with the Ministry of Science and Higher Education of the Russian Federation.

PY - 2025/8

Y1 - 2025/8

N2 - We propose a universal neural network architecture for single-stage multi-class polygonal model generation of anatomical structures from three-dimensional medical images. The key component of the architecture is a trainable affine module that dynamically positions and scales the initial meshes of anatomical structures. This eliminates the need for manual template preparation and reduces the number of self-intersections in the resulting meshes. The effectiveness of the proposed approach has been confirmed on the CHAOS and MMWHS datasets. On CHAOS, an average Dice score of 0.958 is achieved with an ASSD of 1.399 mm, and self-intersections are observed in only 2 out of 20 generated surfaces. On MMWHS, the average Dice score across heart structures is approximately 0.9, and the proportion of self-intersecting edges is comparable to or lower than in the best available methods. Overall, the results demonstrate an accuracy level comparable to modern standards, while producing meshes with significantly cleaner topology. Ablation analysis also confirms the importance of the affine module for generating topologically correct polygonal models.

AB - We propose a universal neural network architecture for single-stage multi-class polygonal model generation of anatomical structures from three-dimensional medical images. The key component of the architecture is a trainable affine module that dynamically positions and scales the initial meshes of anatomical structures. This eliminates the need for manual template preparation and reduces the number of self-intersections in the resulting meshes. The effectiveness of the proposed approach has been confirmed on the CHAOS and MMWHS datasets. On CHAOS, an average Dice score of 0.958 is achieved with an ASSD of 1.399 mm, and self-intersections are observed in only 2 out of 20 generated surfaces. On MMWHS, the average Dice score across heart structures is approximately 0.9, and the proportion of self-intersecting edges is comparable to or lower than in the best available methods. Overall, the results demonstrate an accuracy level comparable to modern standards, while producing meshes with significantly cleaner topology. Ablation analysis also confirms the importance of the affine module for generating topologically correct polygonal models.

KW - МУЛЬТИКЛАССОВАЯ ГЕНЕРАЦИЯ ПОВЕРХНОСТЕЙ

KW - ПОЛИГОНАЛЬНЫЕ СЕТКИ

KW - АФФИННОЕ ВЫРАВНИВАНИЕ

KW - НЕЙРОННЫЕ СЕТИ

KW - multi-class surface generation

KW - polygonal meshes

KW - affine alignment

KW - neural networks

UR - https://www.scopus.com/pages/publications/105031715542

UR - https://www.mendeley.com/catalogue/d37e0b5e-fefc-3c81-ada3-561e5018fdc6/

U2 - 10.1134/S1064562425700231

DO - 10.1134/S1064562425700231

M3 - Article

VL - 112

SP - 332

EP - 341

JO - Doklady Mathematics

JF - Doklady Mathematics

SN - 1064-5624

IS - 1

ER -