Characterizing colon cancer stages through optical polarimetry-assisted digital staining

Tissue polarimetry has been gaining importance in extracting useful diagnostic information from the structural attributes of tissues, which vary in response to the tissue health status and hence find great potential in cancer diagnosis. However, the complexities associated with cancer make it challenging to isolate the characteristic changes as the tumor progresses using polarimetry. This study attempts to experimentally characterize the polarimetric behavior in colon cancer associated with various stages of development. Bulk and unstained sections of normal and tumor colon tissue were imaged in the reflection and transmission polarimetry configurations at low and high imaging resolutions using an in-house developed Mueller polarimeter. Through this study, we observed that the information about the major contributors of scattering in colon tissue, manifesting in depolarization and retardance, can be obtained from the bulk tissue and unstained sections. These parameters aid in characterizing the polarimetric changes as the colon tumor progresses. While the unstained colon section best indicated the depolarization contrast between normal and tumor, the contrast through the retardance parameter was more pronounced in the bulk colon tissue. The results suggest that the polarimetric "digitally stained" images obtained by Mueller polarimetry are comparable with the bulk tissue counterparts, making it useful for characterizing colon cancer tissues across different stages of development.

Automatic normalized digital color staining in the recognition of abnormal blood cells using generative adversarial networks

BACKGROUND AND OBJECTIVES

Combining knowledge of clinical pathologists and deep learning models is a growing trend in morphological analysis of cells circulating in blood to add objectivity, accuracy, and speed in diagnosing hematological and non-hematological diseases. However, the variability in staining protocols across different laboratories can affect the color of images and performance of automatic recognition models. The objective of this work is to develop, train and evaluate a new system for the normalization of color staining of peripheral blood cell images, so that it transforms images from different centers to map the color staining of a reference center (RC) while preserving the structural morphological features.

METHODS

The system has two modules, GAN1 and GAN2. GAN1 uses the PIX2PIX technique to fade original color images to an adaptive gray, while GAN2 transforms them into RGB normalized images. Both GANs have a similar structure, where the generator is a U-NET convolutional neural network with ResNet and the discriminator is a classifier with ResNet34 structure. Digitally stained images were evaluated using GAN metrics and histograms to assess the ability to modify color without altering cell morphology. The system was also evaluated as a pre-processing tool before cells undergo a classification process. For this purpose, a CNN classifier was designed for three classes: abnormal lymphocytes, blasts and reactive lymphocytes.

RESULTS

Training of all GANs and the classifier was performed using RC images, while evaluations were conducted using images from four other centers. Classification tests were performed before and after applying the stain normalization system. The overall accuracy reached a similar value around 96% in both cases for the RC images, indicating the neutrality of the normalization model for the reference images. On the contrary, it was a significant improvement in the classification performance when applying the stain normalization to the other centers. Reactive lymphocytes were the most sensitive to stain normalization, with true positive rates (TPR) increasing from 46.3% - 66% for the original images to 81.2% - 97.2% after digital staining. Abnormal lymphocytes TPR ranged from 31.9% - 95.7% with original images to 83% - 100% with digitally stained images. Blast class showed TPR ranges of 90.3% - 94.4% and 94.4% - 100%, for original and stained images, respectively.

CONCLUSIONS

The proposed GAN-based normalization staining approach improves the performance of classifiers with multicenter data sets by generating digitally stained images with a quality similar to the original images and adaptability to a reference staining standard. The system requires low computation cost and can help improve the performance of automatic recognition models in clinical settings.

Comparison of deep learning models for digital H&E staining from unpaired label-free multispectral microscopy images

BACKGROUND AND OBJECTIVE

This paper presents the quantitative comparison of three generative models of digital staining, also known as virtual staining, in H&E modality (i.e., Hematoxylin and Eosin) that are applied to 5 types of breast tissue. Moreover, a qualitative evaluation of the results achieved with the best model was carried out. This process is based on images of samples without staining captured by a multispectral microscope with previous dimensional reduction to three channels in the RGB range.

METHODS

The models compared are based on conditional GAN (pix2pix) which uses images aligned with/without staining, and two models that do not require image alignment, Cycle GAN (cycleGAN) and contrastive learning-based model (CUT). These models are compared based on the structural similarity and chromatic discrepancy between samples with chemical staining and their corresponding ones with digital staining. The correspondence between images is achieved after the chemical staining images are subjected to digital unstaining by means of a model obtained to guarantee the cyclic consistency of the generative models.

RESULTS

The comparison of the three models corroborates the visual evaluation of the results showing the superiority of cycleGAN both for its larger structural similarity with respect to chemical staining (mean value of SSIM ∼ 0.95) and lower chromatic discrepancy (10%). To this end, quantization and calculation of EMD (Earth Mover's Distance) between clusters is used. In addition, quality evaluation through subjective psychophysical tests with three experts was carried out to evaluate quality of the results with the best model (cycleGAN).

CONCLUSIONS

The results can be satisfactorily evaluated by metrics that use as reference image a chemically stained sample and the digital staining images of the reference sample with prior digital unstaining. These metrics demonstrate that generative staining models that guarantee cyclic consistency provide the closest results to chemical H&E staining that also is consistent with the result of qualitative evaluation by experts.

Diagnostic Accuracy of Digital Staining ex vivo Confocal Microscopy for Diagnosing and Subtyping Basal Cell Carcinoma in Fresh Pretherapeutic Punch Biopsies: A Monocentric Prospective Study

BACKGROUND

Ex vivo confocal microscopy using fusion mode and digital staining (EVCM) scans unfixed fresh tissue and produces rapidly digitally stained images of very similar quality to classical pathology. We investigated whether EVCM could represent an alternative to the standard histological examination of the pretherapeutic basal cell carcinoma (BCC) punch biopsies.

OBJECTIVES

The objective of the study was to assess diagnostic accuracy of EVCM versus traditional histopathological examination for diagnosing and subtyping clinically suspicious lesions of BCC in 3-mm fresh and nonfixed punch biopsies.

METHODS

In this prospective monocentric observational study, patients with clinically suspected BCC were consecutively enrolled. Punch biopsies were imaged using EVCM and subsequently processed for standard histologic examination (gold standard). EVCM images were examined by a dermatopathologist blinded to clinical aspect of the lesion and histopathological results. Concordance between the EVCM and histology analysis was calculated with Cohen's kappa (κ) statistic.

RESULTS

Sixty-six patients were recruited, and 106 biopsies were analyzed. EVCM correctly diagnosed 70/73 BCCs and 31/33 non-BCC lesions, corresponding to a sensitivity of 96% and a specificity of 94% (positive predictive value = 97%, negative predictive value = 91%). The EVCM assessment led to over-staging and under-staging of BCC subtypes in 5% and 11% of cases, respectively. It led to over-staging and under-staging of BCC depths in 5% and 15%, respectively. The kappa coefficient for concordance was 0.78 (95% confidence interval [CI]: 0.69-0.88) when considering BCC subtypes and 0.81 (95% CI: 0.72-0.90) when considering BCC depths.

CONCLUSIONS

These results render EVCM as a promising option for "real-time" pretreatment evaluation of clinically suspected BCC lesions. Further larger randomized studies are needed to assess the efficiency of EVCM versus standard care in patients with clinically suspected BCC.

Assessment of tissue pathology using optical polarimetry

Optical polarimetry have been extensively used for the non-invasive assessment of biological tissues. However, the knowledge regarding differences in polarimetric signatures of different tissue pathologies is very scattered, confounding the deduction of a global trend of the polarimetric variables for healthy and pathological tissues. The purpose of this study was to bridge this gap. We conducted a rigorous online survey to collect all published studies that report the two most common polarimetric variables (i.e., depolarization and retardance) for any type of tissue pathology. A total of 101 studies describing the polarimetric assessment of tissues were collected, wherein 253 (i.e., n_human = 149, n_animal = 104) different type of tissues were optically characterized. Most tissue samples (172/253) were investigated in ex vivo settings. The data showed 32 different types of tissues pathologies, where the most common pathology was cancer and its subtypes. The skin tissues were the most frequently explored tissues, followed by tissue samples from breast, colon, liver, and cervix. Although differences in polarimetric signatures of different tissue pathologies were summarized from the included studies, generalization of the results was hindered by the presentation of polarimetric data in a non-uniform format. The analyses presented in this study may provide an important reference for future polarimetric studies that conduct optical assessment of tissues at greater depth, particularly in the context of optical biopsy/digital staining.