Spatial resolution requirements for acquisition of the virtual screening slide for digital whole-specimen breast histopathology

Comparison of four different displays for identification of select pathologic features extracted from whole slide images of surgical pathology cases

BACKGROUND

The establishment of minimum standards for display selection for the whole slide image (WSI) interpretation has not been fully defined. Recently, pathologists have increasingly preferred using remote displays for clinical diagnostics. Our study aims to assess and compare the performance of three fixed work displays and one remote personal display in accurately identifying ten selected pathologic features integrated into WSIs.

DESIGN

Hematoxylin and eosin-stained glass slides were digitized using Philips scanners. Seven practicing pathologists and three residents reviewed ninety WSIs to identify ten pathologic features using the LG, Dell, and Samsung and an optional consumer-grade display. Ten pathologic features included eosinophils, neutrophils, plasma cells, granulomas, necrosis, mucin, hemosiderin, crystals, nucleoli, and mitoses.

RESULTS

The accuracy of the identification of ten features on different types of displays did not significantly differ among the three types of "fixed" workplace displays. The highest accuracy was observed for the identification of neutrophils, eosinophils, plasma cells, granuloma, and mucin. On the other hand, a lower accuracy was observed for the identification of crystals, mitoses, necrosis, hemosiderin, and nucleoli. Participant pathologists and residents preferred the use of larger displays (>30″) with a higher pixel count, resolution, and luminance.

CONCLUSION

Most features can be identified using any display. However, certain features posed more challenges across the three fixed display types. Furthermore, the use of a remote personal consumer-grade display chosen according to the pathologists' preference showed similar feature identification accuracy. Several factors of display characteristics seemed to influence pathologists' display preferences such as the display size, color, contrast ratio, pixel count, and luminance calibration. This study supports the use of standard "unlocked" vendor-agnostic displays for clinical digital pathology workflow rather than purchasing "locked" and more expensive displays that are part of a digital pathology system.

Histo-genomics: digital pathology at the forefront of precision medicine

The toughest challenge OMICs face is that they provide extremely high molecular resolution but poor spatial information. Understanding the cellular/histological context of the overwhelming genetic data is critical for a full understanding of the clinical behavior of a malignant tumor. Digital pathology can add an extra layer of information to help visualize in a spatial and microenvironmental context the molecular information of cancer. Thus, histo-genomics provide a unique chance for data integration. In the era of a precision medicine, a four-dimensional (4D) (temporal/spatial) analysis of cancer aided by digital pathology can be a critical step to understand the evolution/progression of different cancers and consequently tailor individual treatment plans. For instance, the integration of molecular biomarkers expression into a three-dimensional (3D) image of a digitally scanned tumor can offer a better understanding of its subtype, behavior, host immune response and prognosis. Using advanced digital image analysis, a larger spectrum of parameters can be analyzed as potential predictors of clinical behavior. Correlation between morphological features and host immune response can be also performed with therapeutic implications. Radio-histomics, or the interface of radiological images and histology is another emerging exciting field which encompasses the integration of radiological imaging with digital pathological images, genomics, and clinical data to portray a more holistic approach to understating and treating disease. These advances in digital slide scanning are not without technical challenges, which will be addressed carefully in this review with quick peek at its future.

Three-dimensional tumor visualization of invasive breast carcinomas using whole-mount serial section histopathology: implications for tumor size assessment

PURPOSE

Linear tumor size (T-size) estimated with conventional histology informs breast cancer management. Previously we demonstrated significant differences in margin and focality estimates using conventional histology versus digital whole-mount serial sections (WMSS). Using WMSS we can measure T-size or volume. Here, we compare WMSS T-size with volume, and with T-size measured conventionally. We also compare the ellipsoid model for calculating tumor volume to direct, WMSS measurement.

METHODS

Two pathologists contoured regions of invasive carcinoma and measured T-size from both WMSS and (simulated) conventional sections in 55 consecutive lumpectomy specimens. Volume was measured directly from the contours. Measurements were compared using the paired t-test or Spearman's rank-order correlation. A five-point 'border index' was devised and assigned to each case to parametrize tumor shape considering 'compactness' or cellularity. Tumor volumes calculated assuming ellipsoid geometry were compared with direct, WMSS measurements.

RESULTS

WMSS reported significantly larger T-size than conventional histology in the majority of cases [61.8%, 34/55; means = (2.34 cm; 1.99 cm), p < 0.001], with a 16.4% (9/55) rate of 'upstaging'. The majority of discordances were due to undersampling. T-size and volume were strongly correlated (r = 0.838, p < 0.001). Significantly lower volume was obtained with WMSS versus ellipsoid modeling [means = (1.18 cm³; 1.45 cm³), p < 0.001].

CONCLUSIONS

Significantly larger T-size is measured with WMSS than conventionally, due primarily to undersampling in the latter. Volume and linear size are highly correlated. Diffuse tumors interspersed with normal or non-invasive elements may be sampled less extensively than more localized masses. The ellipsoid model overestimates tumor volume.

Cost-effectiveness analysis of whole-mount pathology processing for patients with early breast cancer undergoing breast conservation

BACKGROUND

Obtaining accurate histopathologic detail for breast lumpectomy specimens is challenging because of sampling and loss of three-dimensional conformational features with conventional processing. The whole-mount (wm) technique is a novel method of serial pathologic sectioning designed to optimize cross-sectional visualization of resected specimens and determination of margin status.

METHODS

Using a Markov chain cohort simulation cost-effectiveness model, we compared conventional processing with wm technique for breast lumpectomies. Cost-effectiveness was evaluated from the perspective of the Canadian health care system and compared using incremental cost-effectiveness ratios (icers) for cost per quality-adjusted life-year (qaly) over a 10-year time horizon. Deterministic and probabilistic sensitivity analyses were performed to test the robustness of the model with willingness-to-pay (wtp) thresholds of $0-$100,000. Costs are reported in adjusted 2014 Canadian dollars, discounted at a rate of 3%.

RESULTS

Compared with conventional processing, wm processing is more costly ($19,989 vs. $18,427) but generates 0.03 more qalys over 10 years. The icer is $45,414, indicating that this additional amount is required for each additional qaly obtained. The model was robust to all variance in parameters, with the prevalence of positive margins accounting for most of the model's variability.

CONCLUSIONS

After a wtp threshold of $45,414, wm processing becomes cost-effective and ultimately generates fewer recurrences and marginally more qalys over time. Excellent baseline outcomes for the current treatment of breast cancer mean that incremental differences in survival are small. However, the overall benefit of the wm technique should be considered in the context of achieving improved accuracy and not just enhancements in clinical effectiveness.

Whole slide imaging diagnostic concordance with light microscopy for breast needle biopsies

This study investigated the diagnostic accuracy of whole slide imaging (WSI) in breast needle biopsy diagnosis in comparison with standard light microscopy (LM). The study examined the effects of image capture magnification and computer monitor quality on diagnostic concordance of WSI and LM. Four pathologists rendered diagnoses using WSI to examine 85 breast biopsies (92 parts; 786 slides) consisting of benign and malignant cases. Each WSI case was evaluated using images captured at either ×20 or ×40 magnifications and viewed using a Digital Imaging and Communication in Medicine (DICOM) grade, color-calibrated monitor or a standard, desktop liquid-crystal display (LCD) monitor. For each combination, the WSI result was compared with the original, LM diagnosis. The overall concordance rate observed between WSI and LM was 97.1% (95% confidence intervals [CI]: 94.3%-98.5%). After a washout period, all cases were reviewed a second time by each pathologist after using LM, and the second LM diagnosis was compared with the WSI diagnosis rendered by the same pathologist. Intraobserver concordance between WSI and LM was 95.4% (95% CI: 92.2%-97.4%). The second LM diagnoses were also compared with the original LM diagnoses, and the observed interobserver LM concordance rate was 97.3% (95% CI: 93.1%-99.0%). The study data demonstrated that breast needle biopsy diagnoses rendered by WSI were equivalent to diagnoses rendered by LM. No diagnostic differences were detected between the underlying viewing system parameters of monitor quality and image capture resolution. The results of this study demonstrated that WSI can be effectively used in subspecialty diagnostic cases where a minimum amount of tissue is available.

3D Pathology Volumetric Technique: A Method for Calculating Breast Tumour Volume from Whole-Mount Serial Section Images

Tumour size, most commonly measured by maximum linear extent, remains a strong predictor of survival in breast cancer. Tumour volume, proportional to the number of tumour cells, may be a more accurate surrogate for size. We describe a novel “3D pathology volumetric technique” for lumpectomies and compare it with 2D measurements. Volume renderings and total tumour volume are computed from digitized whole-mount serial sections using custom software tools. Results are presented for two lumpectomy specimens selected for tumour features which may challenge accurate measurement of tumour burden with conventional, sampling-based pathology: (1) an infiltrative pattern admixed with normal breast elements; (2) a localized invasive mass separated from the in situ component by benign tissue. Spatial relationships between key features (tumour foci, close or involved margins) are clearly visualized in volume renderings. Invasive tumour burden can be underestimated using conventional pathology, compared to the volumetric technique (infiltrative pattern: 30% underestimation; localized mass: 3% underestimation for invasive tumour, 44% for in situ component). Tumour volume approximated from 2D measurements (i.e., maximum linear extent), assuming elliptical geometry, was seen to overestimate volume compared to the 3D volumetric calculation (by a factor of 7x for the infiltrative pattern; 1.5x for the localized invasive mass).

Cytologic evaluation of image-guided fine needle aspiration biopsies via robotic microscopy: A validation study

Background:

This study carried out was to assess the feasibility of using robotic microscopy (RM) for cytologic evaluation of direct smears from fine needle aspiration biopsy (FNAB).

Methods:

Three board-certified cytopathologists reviewed representative direct smears from 40 image-guided FNABs using RM and subsequently re-reviewed the same smears using conventional microscopy. Adequacy of the smears and cytologic diagnosis, as determined using the two approaches, were compared for each individual cytopathologist (intraobserver) and between the three cytopathologists (interobserver). The intraobserver and interobserver discrepancies were analyzed and discussed in a follow-up consensus conference.

Results:

For assessment of adequacy, there were high concordance rates (intraobserver: 92.5–97.5%; interobserver: 90–92.5%), with a few discrepancies involving distinctions between suboptimal and satisfactory smears. Analysis of diagnostic interpretations showed correct classification of 92.5–95% (intraobserver) or 90–92.5% (interobserver) of benign and malignant cases combined, with the discrepancies being between benign and atypical cells in the benign group, and between suspicious and malignant in the malignant group. Within the malignant group, 94% of cases were accurately subclassified via RM. The quality of images viewed by using RM was rated adequate (fair or good) for 95% of the slides.

Conclusions:

The results demonstrate that cytologic evaluation of direct smears from FNABs using RM is feasible. Problems encountered included the longer times needed to evaluate cases with thick, bloody smears and/or low numbers of diagnostic cells, and difficulties in recognizing neuroendocrine differentiation and mimics of hepatocellular carcinoma.

A web-based solution for viewing large-sized microscopic images

This paper demonstrates a pure web-based solution enabling the presentation of scanned pathologic microscopic images on the web. For each slide, an entire specimen is scanned, and a high-resolution digital image (in the order of giga-pixels) is reconstructed. These huge images are then tiled into many 256 x 256-pixel blocks with different resolutions, and information about the blocks of each scanned slide is included in an extensible markup language metafile. Based on the data, a virtual microscopy system is created for viewing the scanned pathologic slides on web. The functionalities (changing viewing resolution, location adjustment, and multimedia annotation presentation) of our virtual slide viewing system are accomplished using pure hypertext markup language (HTML) and JavaScript. We show that there is no need to add plug-in components to browsers in order to handle virtual slides on the web. In a heterogeneous healthcare environment, methods using pure HTML and JavaScript to deal with pathologic content are more appropriate than using proprietary technologies supported only by specific browsers.

[On line digital microscopy in 2007: One technology, many uses]

Digital microscopy enables several observers to look at any field of one microscopical section, at any magnification, through an Internet connexion. An overview of the systems used to digitize microscopy slides and to put them on line is presented. This technique is already used in many fields of pathology, for teaching, research and, to a lesser extent, diagnostic purposes. Several examples are given in this review, some of them with a true evaluation process, and strong points and weaker points are addressed. While conventional microscopy remains the keystone method in 2007 and for the coming years, it is also obvious that digital microscopy will be playing an increasing role. It is our task to make it evolve according to our needs.