An increasing volume of prostate biopsies and a worldwide shortage of urological pathologists puts a strain on pathology departments. Additionally, the high intra-observer and inter-observer ...variability in grading can result in overtreatment and undertreatment of prostate cancer. To alleviate these problems, we aimed to develop an artificial intelligence (AI) system with clinically acceptable accuracy for prostate cancer detection, localisation, and Gleason grading.
We digitised 6682 slides from needle core biopsies from 976 randomly selected participants aged 50–69 in the Swedish prospective and population-based STHLM3 diagnostic study done between May 28, 2012, and Dec 30, 2014 (ISRCTN84445406), and another 271 from 93 men from outside the study. The resulting images were used to train deep neural networks for assessment of prostate biopsies. The networks were evaluated by predicting the presence, extent, and Gleason grade of malignant tissue for an independent test dataset comprising 1631 biopsies from 246 men from STHLM3 and an external validation dataset of 330 biopsies from 73 men. We also evaluated grading performance on 87 biopsies individually graded by 23 experienced urological pathologists from the International Society of Urological Pathology. We assessed discriminatory performance by receiver operating characteristics and tumour extent predictions by correlating predicted cancer length against measurements by the reporting pathologist. We quantified the concordance between grades assigned by the AI system and the expert urological pathologists using Cohen's kappa.
The AI achieved an area under the receiver operating characteristics curve of 0·997 (95% CI 0·994–0·999) for distinguishing between benign (n=910) and malignant (n=721) biopsy cores on the independent test dataset and 0·986 (0·972–0·996) on the external validation dataset (benign n=108, malignant n=222). The correlation between cancer length predicted by the AI and assigned by the reporting pathologist was 0·96 (95% CI 0·95–0·97) for the independent test dataset and 0·87 (0·84–0·90) for the external validation dataset. For assigning Gleason grades, the AI achieved a mean pairwise kappa of 0·62, which was within the range of the corresponding values for the expert pathologists (0·60–0·73).
An AI system can be trained to detect and grade cancer in prostate needle biopsy samples at a ranking comparable to that of international experts in prostate pathology. Clinical application could reduce pathology workload by reducing the assessment of benign biopsies and by automating the task of measuring cancer length in positive biopsy cores. An AI system with expert-level grading performance might contribute a second opinion, aid in standardising grading, and provide pathology expertise in parts of the world where it does not exist.
Swedish Research Council, Swedish Cancer Society, Swedish eScience Research Center, EIT Health.
Deep Learning in Image Cytometry: A Review Gupta, Anindya; Harrison, Philip J.; Wieslander, Håkan ...
Cytometry. Part A,
April 2019, Volume:
95, Issue:
4
Journal Article
Virtual reality (VR) enables data visualization in an immersive and engaging manner, and it can be used for creating ways to explore scientific data. Here, we use VR for visualization of 3D histology ...data, creating a novel interface for digital pathology to aid cancer research.
Our contribution includes 3D modeling of a whole organ and embedded objects of interest, fusing the models with associated quantitative features and full resolution serial section patches, and implementing the virtual reality application. Our VR application is multi-scale in nature, covering two object levels representing different ranges of detail, namely organ level and sub-organ level. In addition, the application includes several data layers, including the measured histology image layer and multiple representations of quantitative features computed from the histology.
In our interactive VR application, the user can set visualization properties, select different samples and features, and interact with various objects, which is not possible in the traditional 2D-image view used in digital pathology. In this work, we used whole mouse prostates (organ level) with prostate cancer tumors (sub-organ objects of interest) as example cases, and included quantitative histological features relevant for tumor biology in the VR model.
Our application enables a novel way for exploration of high-resolution, multidimensional data for biomedical research purposes, and can also be used in teaching and researcher training. Due to automated processing of the histology data, our application can be easily adopted to visualize other organs and pathologies from various origins.
Unreliable predictions can occur when an artificial intelligence (AI) system is presented with data it has not been exposed to during training. We demonstrate the use of conformal prediction to ...detect unreliable predictions, using histopathological diagnosis and grading of prostate biopsies as example. We digitized 7788 prostate biopsies from 1192 men in the STHLM3 diagnostic study, used for training, and 3059 biopsies from 676 men used for testing. With conformal prediction, 1 in 794 (0.1%) predictions is incorrect for cancer diagnosis (compared to 14 errors 2% without conformal prediction) while 175 (22%) of the predictions are flagged as unreliable when the AI-system is presented with new data from the same lab and scanner that it was trained on. Conformal prediction could with small samples (N = 49 for external scanner, N = 10 for external lab and scanner, and N = 12 for external lab, scanner and pathology assessment) detect systematic differences in external data leading to worse predictive performance. The AI-system with conformal prediction commits 3 (2%) errors for cancer detection in cases of atypical prostate tissue compared to 44 (25%) without conformal prediction, while the system flags 143 (80%) unreliable predictions. We conclude that conformal prediction can increase patient safety of AI-systems.
Stochastic optical reconstruction microscopy (STORM), a super-resolution microscopy technique based on single-molecule localizations, has become popular to characterize sub-diffraction limit targets. ...However, due to lengthy image acquisition, STORM recordings are prone to sample drift. Existing cross-correlation or fiducial marker-based algorithms allow correcting the drift within each channel, but misalignment between channels remains due to interchannel drift accumulating during sequential channel acquisition. This is a major drawback in multi-color STORM, a technique of utmost importance for the characterization of various biological interactions.
We developed RegiSTORM, a software for reducing channel misalignment by accurately registering STORM channels utilizing fiducial markers in the sample. RegiSTORM identifies fiducials from the STORM localization data based on their non-blinking nature and uses them as landmarks for channel registration. We first demonstrated accurate registration on recordings of fiducials only, as evidenced by significantly reduced target registration error with all the tested channel combinations. Next, we validated the performance in a more practically relevant setup on cells multi-stained for tubulin. Finally, we showed that RegiSTORM successfully registers two-color STORM recordings of cargo-loaded lipid nanoparticles without fiducials, demonstrating the broader applicability of this software.
The developed RegiSTORM software was demonstrated to be able to accurately register multiple STORM channels and is freely available as open-source (MIT license) at https://github.com/oystein676/RegiSTORM.git and https://doi.org/10.5281/zenodo.5509861 (archived), and runs as a standalone executable (Windows) or via Python (Mac OS, Linux).
Numerous studies have shown a correlation between perineural invasion (PNI) in prostate biopsies and outcome. The reporting of PNI varies widely in the literature. While the interobserver variability ...of prostate cancer grading has been studied extensively, less is known regarding the reproducibility of PNI. A total of 212 biopsy cores from a population-based screening trial were included in this study (106 with and 106 without PNI according to the original pathology reports). The glass slides were scanned and circulated among four pathologists with a special interest in urological pathology for assessment of PNI. Discordant cases were stained by immunohistochemistry for S-100 protein. PNI was diagnosed by all four observers in 34.0% of cases, while 41.5% were considered to be negative for PNI. In 24.5% of cases, there was a disagreement between the observers. The kappa for interobserver variability was 0.67–0.75 (mean 0.73). The observations from one participant were compared with data from the original reports, and a kappa for intraobserver variability of 0.87 was achieved. Based on immunohistochemical findings among discordant cases, 88.6% had PNI while 11.4% did not. The most common diagnostic pitfall was the presence of bundles of stroma or smooth muscle. It was noted in a few cases that collagenous micronodules could be mistaken for a nerve. The distance between cancer and nerve was another cause of disagreement. Although the results suggest that the reproducibility of PNI may be greater than that of prostate cancer grading, there is still a need for improvement and standardization.
Adipose tissue is an attractive stem cell source for soft and bone tissue engineering applications and stem cell therapies. The adipose-derived stromal/stem cells (ASCs) have a multilineage ...differentiation capacity that is regulated through extracellular signals. The cellular events related to cell adhesion and cytoskeleton have been suggested as central regulators of differentiation fate decision. However, the detailed knowledge of these molecular mechanisms in human ASCs remains limited. This study examined the significance of focal adhesion kinase (FAK), Rho-Rho-associated protein kinase (Rho-ROCK), and their downstream target extracellular signal-regulated kinase 1/2 (ERK1/2) on hASCs differentiation towards osteoblasts and adipocytes. Analyses of osteogenic markers RUNX2A, alkaline phosphatase, and matrix mineralization revealed an essential role of active FAK, ROCK, and ERK1/2 signaling for the osteogenesis of hASCs. Inhibition of these kinases with specific small molecule inhibitors diminished osteogenesis, while inhibition of FAK and ROCK activity led to elevation of adipogenic marker genes AP2 and LEP and lipid accumulation implicating adipogenesis. This denotes to a switch-like function of FAK and ROCK signaling in the osteogenic and adipogenic fates of hASCs. On the contrary, inhibition of ERK1/2 kinase activity deceased adipogenic differentiation, indicating that activation of ERK signaling is required for both adipogenic and osteogenic potential. Our findings highlight the reciprocal role of cell adhesion mechanisms and actin dynamics in regulation of hASC lineage commitment. This study enhances the knowledge of molecular mechanisms dictating hASC differentiation and thus opens possibilities for more efficient control of hASC differentiation.
The International Society of Urological Pathology (ISUP) hosts a reference image database supervised by experts with the purpose of establishing an international standard in prostate cancer grading. ...Here, we aimed to identify areas of grading difficulties and compare the results with those obtained from an artificial intelligence system trained in grading. In a series of 87 needle biopsies of cancers selected to include problematic cases, experts failed to reach a 2/3 consensus in 41.4% (36/87). Among consensus and non-consensus cases, the weighted kappa was 0.77 (range 0.68–0.84) and 0.50 (range 0.40–0.57), respectively. Among the non-consensus cases, four main causes of disagreement were identified: the distinction between Gleason score 3 + 3 with tangential cutting artifacts vs. Gleason score 3 + 4 with poorly formed or fused glands (13 cases), Gleason score 3 + 4 vs. 4 + 3 (7 cases), Gleason score 4 + 3 vs. 4 + 4 (8 cases) and the identification of a small component of Gleason pattern 5 (6 cases). The AI system obtained a weighted kappa value of 0.53 among the non-consensus cases, placing it as the observer with the sixth best reproducibility out of a total of 24. AI may serve as a decision support and decrease inter-observer variability by its ability to make consistent decisions. The grading of these cancer patterns that best predicts outcome and guides treatment warrants further clinical and genetic studies. Results of such investigations should be used to improve calibration of AI systems.
Polydimethylsiloxane (PDMS) is widely used in dynamic biological microfluidic applications. As a highly hydrophobic material, native PDMS does not support cell attachment and culture, especially in ...dynamic conditions. Previous covalent coating methods use glutaraldehyde (GA) which, however, is cytotoxic. This paper introduces a novel and simple method for binding collagen type I covalently on PDMS using ascorbic acid (AA) as a cross-linker instead of GA. We compare the novel method against physisorption and GA cross-linker-based methods. The coatings are characterized by immunostaining, contact angle measurement, atomic force microscopy and infrared spectroscopy, and evaluated in static and stretched human adipose stem cell (hASC) cultures up to 13 days. We found that AA can replace GA as a cross-linker in the covalent coating method and that the coating is durable after sonication and after 6 days of stretching. Furthermore, we show that hASCs attach and proliferate better on AA cross-linked samples compared with physisorbed or GA-based methods. Thus, in this paper, we provide a new PDMS coating method for studying cells, such as hASCs, in static and dynamic conditions. The proposed method is an important step in the development of PDMS-based devices in cell and tissue engineering applications.