TY - JOUR

T1 - Cell Structure Segmentation in TEM Images of Murine Skin Melanoma Cells by Deep Learning Model

AU - Genaev, Mikhail A.

AU - Gogaeva, Izabella S.

AU - Taskaeva, Iuliia S.

AU - Bgatova, Nataliya P.

AU - Kozhekin, Mikhail V.

AU - Komyshev, Evgeniy G.

AU - Afonnikov, Dmitry A.

N1 - Genaev, M.A.; Gogaeva, I.S.; Taskaeva, I.S.; Bgatova, N.P.; Kozhekin, M.V.; Komyshev, E.G.; Afonnikov, D.A. Cell Structure Segmentation in TEM Images of Murine Skin Melanoma Cells by Deep Learning Model. J. Imaging 2026, 12, 215. https://doi.org/10.3390/jimaging12050215 Funding: This work was supported by the funding of the Institute of Cytology and genetics SB RAS budget project no. FWNR-2026-0023

PY - 2026/5/26

Y1 - 2026/5/26

N2 - Mitochondria–endoplasmic reticulum contact sites (MERCs) are known as the specialized areas that are involved in a large number of intracellular signaling pathways that regulate Ca2+ homeostasis, lipid transport, mitochondrial dynamics, cell death, and autophagy. Understanding MERC dynamics has important therapeutic implications in cancer, as these contacts regulate fundamental cellular processes and MERCs represent promising targets for therapeutic interventions aimed at improving cancer treatment outcomes. Despite the accumulated data, the role of MERCs in carcinogenesis still remains unknown; thus, it seems promising to search for new tools facilitating the study of MERCs in tumor cells. The structure of MERCs can be examined in great detail using transmission electron microscopy (TEM). Currently, several hundred TEM images are required to obtain reliable data on these contacts. The speed of data processing can be significantly improved by using fast and accurate image analysis techniques based on deep learning models. In this study, five U-Net models with a ResNet34 encoder network were evaluated, including the basic U-Net-Vanilla architecture as well as models incorporating various attention blocks and blocks capturing multilevel image structure, for the segmentation of mitochondria and the endoplasmic reticulum (ER). The best performance on the test dataset was demonstrated by the U-Net-scSE network, with F1 scores of 0.872 for mitochondria and 0.744 for the ER being achieved. Two models were tested for their ability to leverage pre-training on external datasets (Lucchi++, Kasthuri++, and DeepPi-EM). Additionally, models pre-trained on the CEM500K dataset were evaluated after the parameters had been tuned on the data. It was demonstrated by the results that pre-training or the use of pre-trained networks did not lead to an improvement in the IoU and F1 metrics on the test dataset. Subsequent image analysis was conducted to assess two types of MERCs in the segmented images. Finally, the free and user-friendly UltraNet web server was developed for automated analysis of mitochondria, ER, and MERCs using TEM images.

AB - Mitochondria–endoplasmic reticulum contact sites (MERCs) are known as the specialized areas that are involved in a large number of intracellular signaling pathways that regulate Ca2+ homeostasis, lipid transport, mitochondrial dynamics, cell death, and autophagy. Understanding MERC dynamics has important therapeutic implications in cancer, as these contacts regulate fundamental cellular processes and MERCs represent promising targets for therapeutic interventions aimed at improving cancer treatment outcomes. Despite the accumulated data, the role of MERCs in carcinogenesis still remains unknown; thus, it seems promising to search for new tools facilitating the study of MERCs in tumor cells. The structure of MERCs can be examined in great detail using transmission electron microscopy (TEM). Currently, several hundred TEM images are required to obtain reliable data on these contacts. The speed of data processing can be significantly improved by using fast and accurate image analysis techniques based on deep learning models. In this study, five U-Net models with a ResNet34 encoder network were evaluated, including the basic U-Net-Vanilla architecture as well as models incorporating various attention blocks and blocks capturing multilevel image structure, for the segmentation of mitochondria and the endoplasmic reticulum (ER). The best performance on the test dataset was demonstrated by the U-Net-scSE network, with F1 scores of 0.872 for mitochondria and 0.744 for the ER being achieved. Two models were tested for their ability to leverage pre-training on external datasets (Lucchi++, Kasthuri++, and DeepPi-EM). Additionally, models pre-trained on the CEM500K dataset were evaluated after the parameters had been tuned on the data. It was demonstrated by the results that pre-training or the use of pre-trained networks did not lead to an improvement in the IoU and F1 metrics on the test dataset. Subsequent image analysis was conducted to assess two types of MERCs in the segmented images. Finally, the free and user-friendly UltraNet web server was developed for automated analysis of mitochondria, ER, and MERCs using TEM images.

KW - deep learning

KW - endoplasmic reticulum

KW - image segmentation

KW - mitochondria

KW - mitochondria–endoplasmic reticulum contact sites

KW - neural network

KW - transmission electron microscopy

KW - рансмиссионная электронная микроскопия

KW - сегментация изображений

KW - глубокое обучение

KW - нейронная сеть

KW - митохондрии

KW - эндоплазматический ретикулум

KW - места контакта митохондрий и эндоплазматического ретикулума

UR - https://www.mendeley.com/catalogue/9d89e7d8-74b9-3df9-8e04-e7f1687c31cf/

UR - https://www.scopus.com/pages/publications/105039961046

U2 - 10.3390/jimaging12050215

DO - 10.3390/jimaging12050215

M3 - Article

C2 - 42188252

VL - 12

JO - Journal of Imaging

JF - Journal of Imaging

SN - 2313-433X

IS - 5

M1 - 215

ER -