Normalization of H&E stain by autoencoders with analysis of ensemble learning for histological images

• Grant number: 22/03020-1
• Support Opportunities: Research Grants - Initial Project
• Start date: February 01, 2023
• End date: January 31, 2026
• Field of knowledge: Engineering - Biomedical Engineering
• Principal Investigator: Thaína Aparecida Azevedo Tosta
• Grantee: Thaína Aparecida Azevedo Tosta
• Host Institution: Instituto de Ciência e Tecnologia (ICT). Universidade Federal de São Paulo (UNIFESP). Campus São José dos Campos. São José dos Campos , SP, Brazil
• Associated researchers: Leandro Alves Neves ; Marcelo Zanchetta Do Nascimento ; Paulo Rogério de Faria
• Associated scholarship(s): 25/05775-8 - Deep features obtained through LIME and Grad-CAM: an analysis exploring H&E normalized images, BP.IC; 24/00935-4 - Normalization of H&E stain by autoencoders with analysis of ensemble learning for histological images, BP.JC; 24/00936-0 - Normalization of H&E stain by autoencoders with analysis of ensemble learning for histological images, BP.JC; 23/18404-2 - Computational methods for analyses of public H&E histopathological images: a systematic review, BP.IC; 23/15834-6 - Computational method of spectral matching for color normalization of H&E histological images by autoencoders, BP.MS

Abstract

Cancer diagnosis can be confirmed through computational and visual analyses of histological images. However, color variations of these images can harm computational methods' performance. Normalization techniques can be used for correction of these variations caused by the tissue staining process with Hematoxylin-Eosin stain (H&E), commonly used in real clinical practices. Some challenges on studies of literature still make necessary the development of new proposals for normalization improvement. For this purpose, biological properties of stains and tissues can be used to allow a valid biological interpretation of these proposals, and contextual information must also be considered, due to the spatial dependency of histological structures. Thus, this project presents a proposal for investigating spectral matching methods that promote the integration between these concepts with the use of autoencoders for H&E histological images normalization. This proposal will be evaluated with histological images of different types of cancer with evident color variations, to be mapped and identified by a systematic review. Furthermore, it is expected that the use of this methodology contributes to the processing steps of computer-aided diagnosis systems. This evaluation aims to obtain better results with the normalization use in feature extraction and classification steps of histological images. To do so, this project proposes to employ fractal features with different representations, such as LIME and Grad-CAM. These strategies allow better interpretability of the extracted features, with possible gain on images classification. Besides, these representations will be used by deep learning nets in order to evaluate their performance through different ensemble learning models in the process of image class prediction. (AU)

Scientific publications

(10)

(References retrieved automatically from Web of Science and SciELO through information on FAPESP grants and their corresponding numbers as mentioned in the publications by the authors)

TOSTA, THAINA A. AZEVEDO; FREITAS, ANDRE DIAS; DE FARIA, PAULO ROGERIO; NEVES, LEANDRO ALVES; MARTINS, ALESSANDRO SANTANA; DO NASCIMENTO, MARCELO ZANCHETTA. A stain color normalization with robust dictionary learning for breast cancer histological images processing. Biomedical Signal Processing and Control, v. 85, p. 18-pg., 2023-04-27. (22/03020-1)

MIGUEL, JOAO PEDRO MIRANDA; NEVES, LEANDRO ALVES; MARTINS, ALESSANDRO SANTANA; DO NASCIMENTO, MARCELO ZANCHETTA; TOSTA, THAINA A. AZEVEDO. Analysis of neural networks trained with evolutionary algorithms for the classification of breast cancer histological images. EXPERT SYSTEMS WITH APPLICATIONS, v. 231, p. 14-pg., 2023-06-16. (22/03020-1)

DE OLIVEIRA, CLEBER I.; DO NASCIMENTO, MARCELO Z.; ROBERTO, GUILHERME F.; TOSTA, THAINA A. A.; MARTINS, ALESSANDRO S.; NEVES, LEANDRO A.. Hybrid models for classifying histological images: An association of deep features by transfer learning with ensemble classifier. MULTIMEDIA TOOLS AND APPLICATIONS, v. N/A, p. 24-pg., 2023-08-09. (22/03020-1)

SILVA, ADRIANO B.; ROZENDO, GUILHERME B.; TOSTA, THAINA A. A.; MARTINS, ALESSANDRO S.; LOYOLA, ADRIANO M.; CARDOSO, SERGIO V.; LUMINI, ALESSANDRA; NEVES, LEANDRO A.; DE FARIA, PAULO R.; DO NASCIMEMO, MARCELO Z.; et al. CNN Ensembles for Nuclei Segmentation on Histological Images of OED. 2023 IEEE 36TH INTERNATIONAL SYMPOSIUM ON COMPUTER-BASED MEDICAL SYSTEMS, CBMS, v. N/A, p. 4-pg., 2023-01-01. (22/03020-1)

PEREIRA, DANILO C.; LONGO, LEONARDO C.; TOSTA, THAINA A. A.; MARTINS, ALESSANDRO S.; SILVA, ADRIAN B.; ROZENDO, GUILHERME B.; ROBERTO, GUILHERME F.; LUMINI, ALESSANDRA; NEVES, LEANDRO A.; DO NASCIMENTO, MARCELO Z.; et al. Handcrafted features vs deep-learned features: Hermite Polynomial Classification of Liver Images. 2023 IEEE 36TH INTERNATIONAL SYMPOSIUM ON COMPUTER-BASED MEDICAL SYSTEMS, CBMS, v. N/A, p. 6-pg., 2023-01-01. (22/03020-1)

TOSTA, THAINA A. AZEVEDO; DE FARIA, PAULO ROGERIO; NEVES, LEANDRO ALVES; MARTINS, ALESSANDRO SANTANA; KAUSHAL, CHETNA; DO NASCIMENTO, MARCELO ZANCHETTA. Evaluation of sparsity metrics and evolutionary algorithms applied for normalization of H&E histological images. PATTERN ANALYSIS AND APPLICATIONS, v. 27, n. 1, p. 13-pg., 2024-03-01. (22/03020-1)

ROBERTO, GUILHERME F.; PEREIRA, DANILO C.; MARTINS, ALESSANDRO S.; TOSTA, THAINA A. A.; SOARES, CARLOS; LUMINI, ALESSANDRA; ROZENDO, GUILHERME B.; NEVES, LEANDRO A.; NASCIMENTO, MARCELO Z.. Detection of Covid-19 in Chest X-Ray Images Using Percolation Features and Hermite Polynomial Classification. PROGRESS IN PATTERN RECOGNITION, IMAGE ANALYSIS, COMPUTER VISION, AND APPLICATIONS, CIARP 2023, PT I, v. 14469, p. 15-pg., 2024-01-01. (22/03020-1)

SILVA, ADRIANO B.; TOSTA, THALNA A. A.; NEVES, LEANDRO A.; MARTINS, ALESSANDRO S.; DE FARIA, PAULO R.; DO NASCIMENTO, MARCELO Z.. Ensemble of Semantic Segmentation Models for Oral Epithelial Dysplasia Images. 2024 37TH SIBGRAPI CONFERENCE ON GRAPHICS, PATTERNS AND IMAGES, SIBGRAPI 2024, v. N/A, p. 6-pg., 2024-01-01. (22/03020-1)

ROZENDO, GUILHERME BOTAZZO; GARCIA, BIANCA LANCONI DE OLIVEIRA; BORGUE, VINICIUS AUGUSTO TORELI; LUMINI, ALESSANDRA; TOSTA, THAINA APARECIDA AZEVEDO; DO NASCIMENTO, MARCELO ZANCHETTA; NEVES, LEANDRO ALVES. Data Augmentation in Histopathological Classification: An Analysis Exploring GANs with XAI and Vision Transformers. APPLIED SCIENCES-BASEL, v. 14, n. 18, p. 20-pg., 2024-09-01. (22/03020-1)

TENGUAM, JAQUELINE JUNKO; LONGO, LEONARDO H. DA COSTA; ROBERTO, GUILHERME FREIRE; TOSTA, THAINA A. A.; SILVA, ADRIANO B.; DO NASCIMENTO, MARCELO ZANCHETTA; NEVES, LEANDRO ALVES. Higuchi Fractal Dimension with a multidimensional approach for color images. SOFTWARE IMPACTS, v. 21, p. 6-pg., 2024-08-06. (22/03020-1)