Classification of subtypes including LCNEC in lung cancer biopsy slides using convolutional neural network from scratch.

Identifying the lung carcinoma subtype in small biopsy specimens is an important part of determining a suitable treatment plan but is often challenging without the help of special and/or immunohistochemical stains. Pathology image analysis that tackles this issue would be helpful for diagnoses and subtyping of lung carcinoma. In this study, we developed AI models to classify multinomial patterns of lung carcinoma; ADC, LCNEC, SCC, SCLC, and non-neoplastic lung tissue based on convolutional neural networks (CNN or ConvNet). Four CNNs that were pre-trained using transfer learning and one CNN built from scratch were used to classify patch images from pathology whole-slide images (WSIs). We first evaluated the diagnostic performance of each model in the test sets. The Xception model and the CNN built from scratch both achieved the highest performance with a macro average AUC of 0.90. The CNN built from scratch model obtained a macro average AUC of 0.97 on the dataset of four classes excluding LCNEC, and 0.95 on the dataset of three subtypes of lung carcinomas; NSCLC, SCLC, and non-tumor, respectively. Of particular note is that the relatively simple CNN built from scratch may be an approach for pathological image analysis.

Data-efficient and weakly supervised computational pathology on whole-slide images.

Deep-learning methods for computational pathology require either manual annotation of gigapixel whole-slide images (WSIs) or large datasets of WSIs with slide-level labels and typically suffer from poor domain adaptation and interpretability. Here we report an interpretable weakly supervised deep-learning method for data-efficient WSI processing and learning that only requires slide-level labels. The method, which we named clustering-constrained-attention multiple-instance learning (CLAM), uses attention-based learning to identify subregions of high diagnostic value to accurately classify whole slides and instance-level clustering over the identified representative regions to constrain and refine the feature space. By applying CLAM to the subtyping of renal cell carcinoma and non-small-cell lung cancer as well as the detection of lymph node metastasis, we show that it can be used to localize well-known morphological features on WSIs without the need for spatial labels, that it overperforms standard weakly supervised classification algorithms and that it is adaptable to independent test cohorts, smartphone microscopy and varying tissue content.

Metastasis detection from whole slide images using local features and random forests.

Digital pathology has led to a demand for automated detection of regions of interest, such as cancerous tissue, from scanned whole slide images. With accurate methods using image analysis and machine learning, significant speed-up, and savings in costs through increased throughput in histological assessment could be achieved. This article describes a machine learning approach for detection of cancerous tissue from scanned whole slide images. Our method is based on feature engineering and supervised learning with a random forest model. The features extracted from the whole slide images include several local descriptors related to image texture, spatial structure, and distribution of nuclei. The method was evaluated in breast cancer metastasis detection from lymph node samples. Our results show that the method detects metastatic areas with high accuracy (AUC = 0.97-0.98 for tumor detection within whole image area, AUC = 0.84-0.91 for tumor vs. normal tissue detection) and that the method generalizes well for images from more than one laboratory. Further, the method outputs an interpretable classification model, enabling the linking of individual features to differences between tissue types. © 2017 International Society for Advancement of Cytometry.

Reproducibility of atypia of undetermined significance/follicular lesion of undetermined significance category using the bethesda system for reporting thyroid cytology when reviewing slides from different institutions: A study of interobserver variability among cytopathologists.

BACKGROUND: The Bethesda System for Reporting Thyroid Cytopathology (TBSRTC) offers a six-tiered diagnostic scheme for thyroid Fine Needle Aspiration (FNA): Benign, Atypia of Undetermined Significance/Follicular Lesion of Undetermined Significance (AUS/FLUS), suspicious for follicular neoplasm, suspicious for malignancy, malignant, and unsatisfactory with an aim to standardize diagnostic criteria. Reported rate of AUS/FLUS category in the literature has varied from 3% to 20.5%. METHODS: The aim of this study was to assess interobserver variability among cytopathologists to assess reproducibility of the AUS/FLUS category. Seven cytopathologists brought FNA cases (a mixture of atypical and non-atypical FNA diagnosis) diagnosed using TBSRTC from their respective institutions which were reviewed and diagnosed by the participants. The analysis assessed interobserver variability among 7 cytopathologists and determined characteristics on the slides which were associated with concordance to the institutional diagnosis. RESULTS: Seventy eight of 125 (62.4%) benign cases were classified as benign by the reviewers and 26 (21%) were called AUS/FLUS on review. A third of the AUS/FLUS cases were called benign on review and 28.2% were classified as suspicious for neoplasia/malignancy. Roughly a third each of the suspicious for follicular neoplasm/suspicious for malignancy cases were classified as AUS/FLUS. DISCUSSION: When pathologists from different institutions shared their slides, concordance was high for specimens with adequate cellularity and those that were clearly benign but thresholds varied for the other indeterminate categories. Most definite categorization of the AUS/FLUS category was seen on review. Diagn. Cytopathol. 2017;45:399-405. © 2017 Wiley Periodicals, Inc.

A 30-year, population-based study shows improved management and prognosis of hepatocellular carcinoma.

BACKGROUND & AIMS: Little is known about the impact of changes in the management of hepatocellular carcinoma (HCC) over time. We assessed trends in the pattern of care and in prognosis at a population level. METHODS: Data on diagnostic conditions, treatment, and prognosis from 1976-2005 were collected by the population-based digestive cancer registry of Burgundy (France). A nonconditional logistic regression was used to identify factors associated with treatment for cure. A multivariate relative survival analysis was also performed. RESULTS: The context of HCC diagnosis has changed; the proportion of asymptomatic patients increased from 5.6% (1976-1985) to 37.2% (1996-2005). The proportion of cases diagnosed on the basis of morphologic criteria increased from 14% during 1976-1985 to 35.6% during 1996-2005, whereas histologically verified cases decreased from 62.2% to 41.2% between the same time periods. The proportion of patients who were treated with intent to cure increased from 2.7% (1976-1985) to 19.6% (1996-2005). This increase was associated with improvements in relative survival from 4.7% (1976-1985) to 32.8% (1996-2005) at 1 year and from 1.4% to 10.0% at 5 years. The 5-year relative survival of patients treated with curative intent increased, reaching 46.6% for the 1996-2005 period. In the multivariate relative survival analysis, age, period of diagnosis, clinical presentation, alpha-fetoprotein level, and treatment were independent prognostic factors. CONCLUSIONS: During a 30-year period, there was an increase in the number of HCCs diagnosed in asymptomatic subjects that was associated with the development of new effective therapies; this association might account for improvements in prognosis of patients with HCC.

[Evaluation of a patient cytochemical status using nonparametric statistical methods].

The paper shows limitations of the traditional indices used to evaluate a patient's chemical status and describes methods for assessment of its uniformity, by using nonparametric chi 2 test, identifying patient subgroups, and comparing a few groups. A new procedure is proposed to present cytochemical findings.

Contribution of quantitative lectin histochemistry to characterizing well-differentiated, dedifferentiated and poorly differentiated liposarcomas.

OBJECTIVE: To find new diagnostic markers in the group of lipomatous tumors. STUDY DESIGN: The histochemical lectin staining pattern was characterized in a series of 45 lipomatous lesions, including 10 typical lipomas, 6 atypical lipomas, 8 well-differentiated, 6 myxoid, 5 dedifferentiated and 10 pleomorphic liposarcomas. Three lectins were used-peanut (Arachis hypogaea) agglutinin, which binds to terminal Gal(beta 1,3)GalNAc residues; wheat germ (Triticum vulgare) agglutinin (s-WGA, the succinylated form of WGA), which binds to ((1-4)-D-GlcNAc)n and Neu5NAc residues; and jack bean (Concanavalia ensiformis) agglutinin which binds to alpha-D-Man and alpha-D-Glc residues. Histochemical staining was quantitatively measured by means of a cell image processor. RESULTS: In the case of certain carbohydrate residues, typical lipomas closely resemble atypical lipomas, which in turn closely resemble well-differentiated liposarcomas; typical lipomas differ significantly from well-differentiated liposarcomas. This indicates that atypical lipomas, or at least some of them, could represent a biologic link between typical lipomas and well-differentiated liposarcomas. While well-differentiated and pleomorphic liposarcomas differed significantly from each other, the poorly differentiated component of dedifferentiated liposarcomas included histochemical lectin properties, which were common to both well-differentiated and pleomorphic liposarcomas. CONCLUSION: Some atypical lipomas exhibit glycohistochemical characteristics that are common to those of well-differentiated liposarcoma. The poorly differentiated component of dedifferentiated liposarcomas remains more differentiated in terms of glycohistochemical markers than do poorly differentiated pleomorphic liposarcomas.

Asociación entre tiroiditis crónica y miopatía. Un estudio clínico, electromiográfico, histoquímico, de Inmunofluorescencia y microscopia electrónica / Association between chronic thyroiditis and myopathy. A clinical study, electromyographic, histochemic, of immunofluorescence and electronic microscopy

En el presente trabajo fueron estudiados 10 pacientes con diagnóstico de Tiroiditis Linfocítica Crónica, demostrada por biopsia de la glándula tiroidea. Se les realizó el estudio, haciendo énfasis en la búsqueda de síntomas y signos de afección muscular; asimismo se exploró su condición de función tiroidea por T3 y T4 libres y TSH, gammagrama tiroideo y captación del 131, demostrándose la condición de eutiroidismo en todos ellos. También se determinó anticuerpos antiroideos y enzimas musculares. El estudio electromiográfico de los pacientes fue realizado con fines de comprobar la existecia de miopatías; la cual resultó positiva en el 50%, demostrándose potenciales de baja amplitud y corta duración así como presencia polifásicos frecuentes. La velocidad de conducción fue reportado como normal en todos los casos. Fueron realizados 10 biopsias del músculo cuadriceps femoral, a las mismas se les realizó estudio de microscopia óptica, histoquímica, de inmunofluorescencia y microscopia electrónica. La microscopia óptica de las muestras analizadas demostró varialidad en el tamaño y forma de las fibras musculares, vacuolización e infiltrado mononuclear. La histoquímica reveló una disminución y atrofia importante de las fibras tipo I (de conducción lenta), con predominio de las fibras de conducción rápida, sobre las lentas en una relación de 3:1. La inmunofluorescencia señaló actividad anti-inmunoglobulina y anti-C3 en patrón lineal y granular distribuido sobre el sarcoma y dentro de las fibras. La microscopia electrónica mostró cambios degenerativos y necrosis dentro de la fibra, pliegues del sarcolema, engrosamiento de la membrana basal capilar e infiltrado de células mononucleares y mastocitos. Todo ello, sugiere un proceso de miositis posiblemente desencadenado por un mecanismo inmunológico como causa del daño a la fibra muscular, similiar a otras enfermedades autoalérgicas