TY - GEN
T1 - Learned Image Compression for HE-Stained Histopathological Images via Stain Deconvolution
AU - Fischer, Maximilian
AU - Neher, Peter
AU - Wald, Tassilo
AU - Almeida, Silvia Dias
AU - Xiao, Shuhan
AU - Schüffler, Peter
AU - Braren, Rickmer
AU - Götz, Michael
AU - Muckenhuber, Alexander
AU - Kleesiek, Jens
AU - Nolden, Marco
AU - Maier-Hein, Klaus
N1 - Publisher Copyright:
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.
PY - 2025
Y1 - 2025
N2 - Processing histopathological Whole Slide Images (WSI) leads to massive storage requirements for clinics worldwide. Even after lossy image compression during image acquisition, additional lossy compression is frequently possible without substantially affecting the performance of deep learning-based (DL) downstream tasks. In this paper, we show that the commonly used JPEG algorithm is not best suited for further compression and we propose Stain Quantized Latent Compression (SQLC), a novel DL based histopathology data compression approach. SQLC compresses staining and RGB channels before passing it through a compression autoencoder (CAE) in order to obtain quantized latent representations for maximizing the compression. We show that our approach yields superior performance in a classification downstream task, compared to traditional approaches like JPEG, while image quality metrics like the Multi-Scale Structural Similarity Index (MS-SSIM) is largely preserved. Our method is online available.
AB - Processing histopathological Whole Slide Images (WSI) leads to massive storage requirements for clinics worldwide. Even after lossy image compression during image acquisition, additional lossy compression is frequently possible without substantially affecting the performance of deep learning-based (DL) downstream tasks. In this paper, we show that the commonly used JPEG algorithm is not best suited for further compression and we propose Stain Quantized Latent Compression (SQLC), a novel DL based histopathology data compression approach. SQLC compresses staining and RGB channels before passing it through a compression autoencoder (CAE) in order to obtain quantized latent representations for maximizing the compression. We show that our approach yields superior performance in a classification downstream task, compared to traditional approaches like JPEG, while image quality metrics like the Multi-Scale Structural Similarity Index (MS-SSIM) is largely preserved. Our method is online available.
KW - Deep learning
KW - Lossy Compression
KW - Whole Slide Imaging
UR - http://www.scopus.com/inward/record.url?scp=85218242862&partnerID=8YFLogxK
U2 - 10.1007/978-3-031-77786-8_10
DO - 10.1007/978-3-031-77786-8_10
M3 - Conference contribution
AN - SCOPUS:85218242862
SN - 9783031777851
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 97
EP - 107
BT - Medical Optical Imaging and Virtual Microscopy Image Analysis - Second International Workshop, MOVI 2024, Held in Conjunction with MICCAI 2024, Proceedings
A2 - Huo, Yuankai
A2 - Millis, Bryan A.
A2 - Zhou, Yuyin
A2 - Younis, Khaled
A2 - Wang, Xiao
A2 - Tang, Yucheng
PB - Springer Science and Business Media Deutschland GmbH
T2 - 2nd International Workshop on Medical Optical Imaging and Virtual Microscopy Image Analysis, MOVI 2024, held in conjunction with 26th International Conference on Medical Imaging and Computer-Assisted Intervention, MICCAI 2024
Y2 - 10 October 2024 through 10 October 2024
ER -