Applications of Digital Microscopy and Densely Connected Convolutional Neural Networks for Automated Quantification of Babesia-Infected Erythrocytes.

BACKGROUND: Clinical babesiosis is diagnosed, and parasite burden is determined, by microscopic inspection of a thick or thin Giemsa-stained peripheral blood smear. However, quantitative analysis by manual microscopy is subject to error. As such, methods for the automated measurement of percent parasitemia in digital microscopic images of peripheral blood smears could improve clinical accuracy, relative to the predicate method. METHODS: Individual erythrocyte images were manually labeled as "parasite" or "normal" and were used to train a model for binary image classification. The best model was then used to calculate percent parasitemia from a clinical validation dataset, and values were compared to a clinical reference value. Lastly, model interpretability was examined using an integrated gradient to identify pixels most likely to influence classification decisions. RESULTS: The precision and recall of the model during development testing were 0.92 and 1.00, respectively. In clinical validation, the model returned increasing positive signal with increasing mean reference value. However, there were 2 highly erroneous false positive values returned by the model. Further, the model incorrectly assessed 3 cases well above the clinical threshold of 10%. The integrated gradient suggested potential sources of false positives including rouleaux formations, cell boundaries, and precipitate as deterministic factors in negative erythrocyte images. CONCLUSIONS: While the model demonstrated highly accurate single cell classification and correctly assessed most slides, several false positives were highly incorrect. This project highlights the need for integrated testing of machine learning-based models, even when models in the development phase perform well.

Ultrahigh-speed point scanning two-photon microscopy using high dynamic range silicon photomultipliers.

Conventional two-photon microscopes use photomultiplier tubes, which enable high sensitivity but can detect relatively few photons per second, forcing longer pixel integration times and limiting maximum imaging rates. We introduce novel detection electronics using silicon photomultipliers that greatly extend dynamic range, enabling more than an order of magnitude increased photon detection rate as compared to state-of-the-art photomultiplier tubes. We demonstrate that this capability can dramatically improve both imaging rates and signal-to-noise ratio (SNR) in two-photon microscopy using human surgical specimens. Finally, to enable wider use of more advanced detection technology, we have formed the OpenSiPM project, which aims to provide open source detector designs for high-speed two-photon and confocal microscopy.

Initial Evaluation of Rapid, Direct-to-Digital Prostate Biopsy Pathology.

CONTEXT.­: Pathologist interobserver discordance is significant in grading of prostate cancer, limiting reliability. Diagnostic reproducibility may be improved with digital images, but adoption faces workflow, cost, and quality challenges. A novel digital method using an alternative tissue processing approach and novel laser microscopy system potentially addresses these issues. OBJECTIVE.­: To evaluate the capability of this new method for primary diagnostic interpretation in clinical prostate biopsy specimens. DESIGN.­: Forty patients with a high likelihood of prostate cancer based on magnetic resonance imaging consented to investigational core biopsy. A subset of samples was used for direct comparison of physical slide preparation effects and time-tracking determination with multiphoton microscopy. Twenty samples were processed for diagnostic comparison between multilevel digital slides and subsequently produced physical slides. A reference diagnosis based on all data was established using grade groups. Level of diagnostic match and requests for immunohistochemistry were compared between physical and digital diagnoses. Immunohistochemical staining and length measurements were secondary outcomes. RESULTS.­: Interpretations based on direct multiphoton imaging yielded diagnoses that were at least as accurate as standard histology; cancer diagnosis correlation was 89% (51 of 57) by physical slides and 95% (53 of 56) by multiphoton microscopy. Grade-level concordance was 73% (44 of 60) by either method. Immunohistochemistry for routine prostate cancer-associated markers on these alternatively processed tissues was unaffected. Alternatively processed tissues resulted in longer measured core and cancer lengths, suggestive of improved orientation and visualization. CONCLUSIONS.­: Findings support high potential for complete interpretation of prostate core biopsies using solely multiphoton microscopy of intact specimens, with potential diagnostic benefits as well as reduced processing time and reduced processing complexity.

Regulated necrosis and failed repair in cisplatin-induced chronic kidney disease.

Cisplatin is an effective chemotherapeutic agent, but significant nephrotoxicity limits its clinical use. Despite extensive investigation of the acute cellular and molecular responses to cisplatin, the mechanisms of progression from acute to chronic kidney injury have not been explored. We used functional and morphological metrics to establish a time-point when the transition from acute and reversible kidney injury to chronic and irreparable kidney disease is clearly established. In mice administered 1 or 2 doses of intraperitoneal cisplatin separated by 2 weeks, kidney function returned toward baseline two weeks after the first dose, but failed to return to normal two weeks following a second dose. Multiphoton microscopy revealed increased glomerular epithelial and proximal tubular damage in kidneys exposed to two doses of cisplatin compared with those exposed to a single dose. In contrast, there was no evidence of fibrosis, macrophage invasion, or decrease in endothelial cell mass in chronically diseased kidneys. Pathway analysis of microarray data revealed regulated necrosis as a key determinant in the development of chronic kidney disease after cisplatin administration. Western blot analysis demonstrated activation of proteins involved in necroptosis and increased expression of kidney injury markers, cellular stress response regulators, and upstream activators of regulated necrosis, including Toll-like receptors 2 and 4. These data suggest that unresolved injury and sustained activation of regulated necrosis pathways, rather than fibrosis, promote the progression of cisplatin-induced acute kidney injury to chronic kidney disease.

Very Deep Convolutional Neural Networks for Morphologic Classification of Erythrocytes.

BACKGROUND: Morphologic profiling of the erythrocyte population is a widely used and clinically valuable diagnostic modality, but one that relies on a slow manual process associated with significant labor cost and limited reproducibility. Automated profiling of erythrocytes from digital images by capable machine learning approaches would augment the throughput and value of morphologic analysis. To this end, we sought to evaluate the performance of leading implementation strategies for convolutional neural networks (CNNs) when applied to classification of erythrocytes based on morphology. METHODS: Erythrocytes were manually classified into 1 of 10 classes using a custom-developed Web application. Using recent literature to guide architectural considerations for neural network design, we implemented a "very deep" CNN, consisting of >150 layers, with dense shortcut connections. RESULTS: The final database comprised 3737 labeled cells. Ensemble model predictions on unseen data demonstrated a harmonic mean of recall and precision metrics of 92.70% and 89.39%, respectively. Of the 748 cells in the test set, 23 misclassification errors were made, with a correct classification frequency of 90.60%, represented as a harmonic mean across the 10 morphologic classes. CONCLUSIONS: These findings indicate that erythrocyte morphology profiles could be measured with a high degree of accuracy with "very deep" CNNs. Further, these data support future efforts to expand classes and optimize practical performance in a clinical environment as a prelude to full implementation as a clinical tool.

Multiphoton microscopy with clearing for three dimensional histology of kidney biopsies.

We present a multiphoton microscopy approach with clearing optimized for pathology evaluation producing image quality comparable to traditional histology. Use of benzyl alcohol/benzyl benzoate with 4',6-diamidino-2-phenylindole and eosin in an optimized imaging setup results in optical sections nearly indistinguishable from traditionally-cut sections. Application to human renal tissue demonstrates diagnostic-level image quality can be maintained through 1 millimeter of tissue. Three dimensional perspectives reveal changes of glomerular capsule cells not evident on single sections. Collagen-derived second harmonic generation can be visualized through entire biopsies. Multiphoton microscopy with clearing has potential for increasing the yield of histologic evaluation of biopsy specimens.

Integrated fluorescence correlation spectroscopy device for point-of-care clinical applications.

We describe an optical system which reduces the cost and complexity of fluorescence correlation spectroscopy (FCS), intended to increase the suitability of the technique for clinical use. Integration of the focusing optics and sample chamber into a plastic component produces a design which is simple to align and operate. We validate the system by measurements on fluorescent dye, and compare the results to a commercial instrument. In addition, we demonstrate its application to measurements of concentration and multimerization of the clinically relevant protein von Willebrand factor (vWF) in human plasma.