Pathological Classification Research Articles

PATH-17. MOLECULAR CLASSIFICATION OF H3K27-WILDTYPE SPINAL CORD GLIOMAS: IMPLICATIONS FOR PROGNOSIS AND THERAPEUTICS

Abstract BACKGROUND H3K27-wildtype spinal cord gliomas (H3WT-SCGs) are rare and heterogeneous tumors that present significant challenges in pathological classification. Traditional histologic diagnoses often do not align well with the underlying molecular characteristics of these tumors. METHODS We characterized 55 cases of H3WT-SCGs using a comprehensive approach that included DNA methylation array, whole-exome sequencing, transcriptome sequencing, proteomics, and single-cell RNA sequencing. Unsupervised clustering analysis was performed to identify distinct molecular subtypes. RESULTS The analysis identified seven distinct spinal cord methylation (SCM) clusters: pilocytic astrocytomas (PA-SPINE), PA-immune enriched (PA-IME), diffuse leptomeningeal glioneuronal tumor (DLGNT), subependymoma (SUBEPN-SPINE), glioblastoma-like (GBM-like), and pleomorphic xanthoastrocytoma-like (PXA-like). There was a 61.8% discordance between SCM classes and traditional histologic diagnoses. The new molecular classification provided a better correlation with molecular features and patient prognosis. BRAF fusions were enriched in PA-SPINE and DLGNT (p = 0.04), and DLGNT showed significant chromosomal alterations, including 1p loss (p = 6.5e-08) and 1q gain (p = 3.6e-05). Notably, we identified two distinct classes of PA-SPINE: adult (older than 18 years) and pediatric. Adult PAs exhibited lower frequencies of BRAF fusions (p = 1.4e-03), shorter tumor lengths (p = 3.7e-04), and longer progression-free survival (p = 0.062). Adult PA neoplastic cells were less differentiated, with higher expression of EGFR and neuronal markers, and lower cytokine and chemokine expression compared to pediatric cases. Single-cell analysis revealed reduced infiltration of monocyte-derived macrophages and proliferative myeloid cells in adult cases. Both PA-IME and PA-pediatric clusters were characterized as immune-hot tumors with high immune cell enrichment. CONCLUSION This study defined clinically relevant molecular classes of H3WT-SCGs that better reflect patient prognosis and molecular characteristics compared to traditional histologic diagnoses. The findings suggest potential therapeutic implications, especially in immune-hot tumors like IME and PA-pediatric, which could benefit from targeted immunotherapies.

Immunofluorescence Use and Techniques in Glomerular Diseases – A Review

Background: Immunofluorescence (IF) studies play an essential role in the evaluation of medical renal biopsies. Particularly, in the study of renal glomerular diseases, where it provides fundamental data for the diagnosis, classification, and etiology of the glomerular pathologies. Diverse techniques may be used to optimize the utilization of IF studies, from variations on the test methodologies to expertise on the interpretation of the results and knowledge of potential pitfalls. Summary: This manuscript presents a brief review on the history of IF and its utilization in kidney pathology, followed by a description of the IF methods, including the use of immunofluorescence on paraffin embedded tissue (paraffin IF), and other novel techniques. Guidelines on how to best report IF findings are reviewed, along with a description of antibodies commonly used in glomerular diseases, highlighting their distribution within the normal kidney and potential pitfalls in interpretation. Finally, the use and interpretation of IF is discussed in more detail in individual entities on a range of glomerular diseases. Key Messages: Immunofluorescence is crucial for interpretation of renal biopsies and diagnosis of glomerular diseases. Knowledge of IF techniques, alternative procedures, its use and proper interpretation is essential for optimal utilization of IF in renal pathology, and this review proposes to serve as a simplified and practical guide on this topic.

BIOM-55. SENSITIVITY OF IMMUNOHISTOCHEMISTRY WITH BRAF VE1 ANTIBODY FOR BRAF V600E IN PRIMARY CENTRAL NERVOUS SYSTEM TUMORS

Abstract BACKGROUND BRAF gene encodes for a key protein of the RAS-RAF-MEK-ERK pathway. BRAF alterations are found in a variety of primary central nervous system (CNS) tumors ranging from 3% of glioblastomas to 60% of pleomorphic xanthoastrocytomas. BRAF V600E mutation is the most common BRAF alteration found in CNS malignancies, and it has been successfully targeted with BRAF/MEK inhibitors. Novel BRAF inhibitors, still in development, have shown promising antitumor activity. Next-generation sequencing (NGS) has become a standard test in the era of histo-molecular diagnosis, but it can be unavailable or inaccessible, and results usually take longer. Immunohistochemistry (ICH) with VE1 antibody for BRAF V600E protein has shown high discordant results with sequencing in other tumor types. We aimed to establish the concordance of BRAF VE1 and NGS sequencing in CNS tumors. METHODS Single-center retrospective analysis of patients with BRAF V600E mutant CNS tumors by NGS testing. RESULTS 30 patients with BRAF V600E mutant primary CNS tumors by NGS were included, 18 were male. 15 were classified as IDH-wildtype gliomas, 4 low grade neuroepithelial gliomas, 8 pleomorphic xanthoastrocytoma, 2 pilocytic astrocytomas and 1 papillary craniopharyngioma. IHC testing with BRAF VE1 antibody was done at the initial pathological classification in 23 of the cases being positive in 15 and negative in 8 cases. In 7 cases IHC was not done but has been requested for retrospective analysis and results are pending. Sensitivity for BRAF VE1 IHC was 65%. CONCLUSION When compared to NGS testing, ICH with BRAF VE1 antibody for BRAF V600E in CNS tumors shows modest sensitivity, being a limited tool when results are needed promptly or NGS is inaccessible. Determining specificity of the test will be crucial to establish its reliability for patient selection. Sequencing should be obtained when possible and remains the more reliable method for detecting V600E.

Application of the Gait Kinematics Index in Patients with Cerebral Palsy

Due to the complexity of the medical issues connected with cerebral palsy (CP), the classification of gait pathologies seems rather difficult. The aim of this study was to asses the usefulness of the Gait Kinematics Index (GKI) from a clinical point of view in the population of patients with CP. The assessment of the possibilities of using the GKI in a group of patients with CP was conducted on the basis of the correlation of its results with the Gillette Gait Index (GGI) and Gait Deviation Index (GDI) values. The distribution of the index values was also evaluated with attention paid to the CP types and treatment methods. Analyses were performed on the basis of the gait test results in a group of 56 healthy children and 72 patients with CP. The GKI values for patients with CP were 1.55 ± 0.66, as opposed to 0.77 ± 0.17 for the reference group. A strong linear correlation was found between the values of the GKI and GGI (r = 0.8 ÷ 0.85), as well as between the GKI and GDI (r = −0.89 ÷ 0.9), obtained in children with CP. In addition, significant differences were found between the results obtained in all the groups of children with CP divided by treatment method (rehabilitation, botulinum, rhizotomy, p < 0.05), whereas in the groups of children divided by CP type, significant differences (p < 0.05) were found solely between diplegia and hemiplegia and between hemiplegia and quadriplegia. The results obtained were the same in the case of the GKI, GGI and GDI. To conclude, the results presented in this work confirm the clinical utility of the GKI in the population of patients with CP.

Artificial intelligence-based morphologic classification and molecular characterization of neuroblastic tumors from digital histopathology

A deep learning model using attention-based multiple instance learning (aMIL) and self-supervised learning (SSL) was developed to perform pathologic classification of neuroblastic tumors and assess MYCN-amplification status using H&E-stained whole slide images from the largest reported cohort to date. The model showed promising performance in identifying diagnostic category, grade, mitosis-karyorrhexis index (MKI), and MYCN-amplification with validation on an external test dataset, suggesting potential for AI-assisted neuroblastoma classification.

Exploring the clinical potential of circulating LncRNAs in breast cancer: insights into primary signaling pathways and therapeutic interventions.

Breast cancer (BC) occupies the top spot among women on a global scale. The tumor has a significant degree of heterogeneity, displaying a notable prevalence of medication resistance, recurrence, and metastasis, rendering it one of the most lethal forms of malignant neoplasms. The timely identification, ongoing evaluation of therapeutic interventions, and accurate prediction of outcomes play crucial roles in determining the overall survival rates of women with BC. Nevertheless, the absence of precise biomarkers remains a significant determinant impacting the overall well-being and both the physical and emotional health of BC patients. Long noncoding RNA (lncRNA) exerts regulatory control over several genes and signaling pathways, hence assuming crucial roles in the development of neoplastic growth. Recently, research has indicated that the atypical expression of circulating lncRNAs in various biological bodily fluids has a noteworthy impact on the early detection, pathological categorization, staging, monitoring of therapy outcomes, and evaluation of prognosis in cases of BC. This article aims to assess the potential clinical utility of circulating lncRNAs in the context of BC focusing on specific primary signaling pathways; Wnt/β-catenin, Notch, TGF-β, and hedgehog (Hh), in addition to some therapeutic interventions.

The clinical pathway in China county for lung cancer diagnosis and treatment (2024 edition)

Standardizing the diagnosis and treatment of lung cancer is a key measure to improve the survival rate of lung cancer patients and reduce the mortality rate. However, county hospitals generally face the problem of inaccessibility to advanced diagnostic and treatment technologies. Therefore, when developing quality control standards and clinical diagnosis and treatment specifications, it is necessary to combine the actual situation of county hospitals and formulate specific recommendations. The recommendations of treatment measures also need to consider the approval status of indications and whether it is included in the National Reimbursement Drug List (NRDL), to ensure the access to medicines. To address the above issues, based on the existing guidelines at home and abroad and the clinical work characteristics of county hospitals, the " the clinical pathway in China county for lung cancer diagnosis and treatment (2024 edition)" has been updated on the basis of the first edition. This pathway elaborated on the imaging diagnosis, pathological diagnosis, molecular testing, precision medicine, and developed different diagnosis and treatment processes for different types of lung cancer patients. Consistent with the first pathway, this update still divides the recommendations for diagnosis and treatment of clinical scenarios into basic strategies and optional strategies for elaboration. The basic strategies are the standards that county hospitals must meet, while the optional strategies provide more choices for hospitals, which are convenient for county doctors to put into clinical practice. All the recommended diagnostic and treatment plans strictly refer to existing guidelines and consensus. Compared to the first edition, based on the latest high-level evidence-based medicine and the approval status of indications, the pathway has updated the diagnosis and treatment recommendations for lung cancer under different pathological types, TNM classification, and molecular classification in basic and optional strategies.

Identification and validation of novel genes related to immune microenvironment in polycystic ovary syndrome.

Polycystic ovary syndrome (PCOS) is one of the most complicated chronic inflammatory diseases in women of reproductive age and is one of the primary factors responsible for infertility. There is substantial dispute relating to the pathophysiology of PCOS. Consequently, there is a critical need for further research to identify the factors underlying the pathophysiology of PCOS. Three transcriptome profiles of granulosa cells from patients with PCOS and normal controls were obtained from the gene expression integration database. We also obtained relevant microarrays of granulocytes prepared from PCOS patients and normal controls from the gene expression integration database. Then, we used the R package to perform correlations and identify differences between PCOS and normal controls with regard to immune infiltrating cells and functionality. Subsequently, intersecting genes were identified and risk models were constructed. Finally, the results were validated by enzyme linked immunosorbent assay and real-time PCR. We identified 8 genes related to cuproptosis (SLC31A1, PDHB, PDHA1, DLST, DLD, DLAT, DBT, and ATP7A) and 5 genes related to m7G (SNUPN, NUDT16, GEMIN5, DCPS, and EIF4E3) that were associated with immune infiltration. Furthermore, the expression levels of DLAT (P = .049) and NUDT16 (P = .024) differed significantly between the PCOS patients and normal controls, as revealed by multifactorial analysis. Both DLAT and NUDT16 were negatively correlated with immune cell expression and function and expression levels were significantly lower in the PCOS group. Finally, real-time PCR and enzyme linked immunosorbent assay demonstrated that the expression levels of DLAT and NUDT16 were significantly reduced in the granulosa cells of PCOS patients. In conclusion, our findings shed fresh light on the roles of immune infiltration, cuproptosis, and m7G alternations in PCOS. We also provide a reliable biomarker for the pathological classification of PCOS patients.

Research progress of breast pathology image diagnosis based on deep learning

Breast cancer is a malignancy caused by the abnormal proliferation of breast epithelial cells, predominantly affecting female patients, and it is commonly diagnosed using histopathological images. Currently, deep learning techniques have made significant breakthroughs in medical image processing, outperforming traditional detection methods in breast cancer pathology classification tasks. This paper first reviewed the advances in applying deep learning to breast pathology images, focusing on three key areas: multi-scale feature extraction, cellular feature analysis, and classification. Next, it summarized the advantages of multimodal data fusion methods for breast pathology images. Finally, the study discussed the challenges and future prospects of deep learning in breast cancer pathology image diagnosis, providing important guidance for advancing the use of deep learning in breast diagnosis.

Microvascular Inflammation of Kidney Allografts and Clinical Outcomes.

The heterogeneous clinical presentation of graft microvascular inflammation poses a major challenge to successful kidney transplantation. The effect of microvascular inflammation on allograft outcomes is unclear. We conducted a cohort study that included kidney-transplant recipients from more than 30 transplantation centers in Europe and North America who had undergone allograft biopsy between 2004 and 2023. We integrated clinical and pathological data to classify biopsy specimens according to the 2022 Banff Classification of Renal Allograft Pathology, which includes two new diagnostic categories: probable antibody-mediated rejection and microvascular inflammation without evidence of an antibody-mediated response. We then assessed the association between the newly recognized microvascular inflammation phenotypes and allograft survival and disease progression. A total of 16,293 kidney-transplant biopsy specimens from 6798 patients were assessed. We identified the newly recognized microvascular inflammation phenotypes in 788 specimens, of which 641 were previously categorized as specimens with no evidence of rejection. As compared with patients without rejection, the hazard ratio for graft loss was 2.1 (95% confidence interval [CI], 1.5 to 3.1) among patients with microvascular inflammation without evidence of an antibody-mediated response and 2.7 (95% CI, 2.2 to 3.3) among patients with antibody-mediated rejection. Patients with a diagnosis of probable antibody-mediated rejection had a higher risk of graft failure beyond year 5 after biopsy than those without rejection (hazard ratio, 1.7; 95% CI, 0.8 to 3.5). Patients with a diagnosis of either newly recognized microvascular inflammation phenotype had a higher risk of progression of transplant glomerulopathy during follow-up than patients without microvascular inflammation. Microvascular inflammation in kidney allografts includes distinct phenotypes, with various disease progression and allograft outcomes. Our findings support the clinical use of additional rejection phenotypes to standardize diagnostics for kidney allografts. (Funded by OrganX. ClinicalTrials.gov number, NCT06496269.).

FEATURES OF ANTIBACTERIAL AND LYMPHOTROPIC THERAPY IN PATIENTS WITH PLEURAL EMPYEMA

Abstract. Acute empyema belongs to the category of severe surgical pathology. The difficulty of its treatment is caused by a number of objective and subjective reasons, in particular such as wide spread of antibiotic-resistant microflora and population allergy. Carefully collected analysis, patient’s examination, study of lungs function indicators, data of laboratory and X-ray studies, performing of serological, immunological and allergic tests do not always allow to identify the cause of acute empyema and are often not sufficient to identify its etiology. In spite of a great number of proposed methods of acute pleural empyema treatment, study of long-term results of all types of treatment shows that they do not guarantee 100% success. The outlined data indicates the necessity of improvement of etiotropic therapy in patients with pleural empyema. Constant search for new effective methods of treatment of the aimed pathology proves the relevance of the theme. The data outlined in this article indicates the necessity of improvement of etiotropic therapy in patients with pleural empyema.

An evaluation of a virtual musculoskeletal podiatry service implemented to address prolonged National Health Service waiting times.

The COVID-19 pandemic had a substantial impact on healthcare systems globally, particularly in the public sector. To address the challenges posed by the pandemic, musculoskeletal (MSK) healthcare providers had to rapidly adopt virtual platforms for delivering care, representing a major shift in how healthcare was delivered. This manuscript aims to retrospectively evaluate a virtual MSK podiatry service offered by a private provider under a National Health Service commission, in terms of patient access, waiting times and patient-reported pain. This service was developed and implemented in response to the COVID-19 pandemic and the extended waiting times. A retrospective clinical service evaluation was conducted on MSK podiatry services delivered via telephone or virtual consultations. The evaluation covered a cohort of 574 referred patients over a 19-month period (July 2021 to January 2023). It analysed demographic data, initial and final visual analogue pain scores, pathology categories, orthoses prescriptions and exercise rehabilitation plans. Data from a total of 492 patients (male=152 and female=340) were analysed, with 82 patients excluded for non-attendance. The average waiting time from referral-to-first appointment and referral-to-discharge was 35 and 91days, respectively. Results showed statistically significant improvement (p<0.001) in the mean visual analogue scale when patients received orthoses with and without a rehabilitation plan (4.12±2.55 and 3.33±2.88, respectively). Most patients (61.5%) were aged 40-69, with "foot pain" being the main reported pathology category. Patients had an average of two appointments. 56.5% of patients remained virtual throughout their journey and were successfully discharged to self-management. 43.9% were discharged to other face-to face services. The study provided evidence that the virtual MSK podiatry service achieved a statistically significant reduction in patient-reported pain for various pathologies with reasonable waiting times. The service delivered favourable outcomes and complemented traditional services at a time with limited access due to the COVID-19 pandemic.

The clinical significance of elastic lamina invasion in patients with pStage II colorectal cancer: a notable prognostic indicator

BackgroundSome colorectal cancers (CRCs) are clinically diagnosed as cT4a with serosal invasion (SI). However, the cT4a is most often underdiagnosed pathologically as pT3 without SI by hematoxylin–eosin (H&E) staining alone. Using Elastica van Gieson (EVG) staining, some pT3 tumors invade the elastic lamina (EL), which extends just below the serosal layer. Recently, EL invasion (ELI) has been described as a poor prognostic factor for disease-free survival (DFS) and overall survival (OS) in patients with pStage II CRC. However, its clinicopathological significance remains unclear due to the limited number of studies and poor understanding of ELI.ObjectiveThis study investigated the association between the ELI and patient prognosis.MethodsAfter 1982, pathological diagnosis was routinely performed using H&E and EVG staining methods, and long-term follow up was performed until 2016. All clinicopathological features including ELI were prospectively registered into our computer and 569 patients with pStage II CRC were collected from the database. Based on the ELI status, pT3 was divided into three pathological categories: pT3ELI − was defined as pT3a, pT3ELI + as pT3b and unidentified EL (pT3EL −) as pT3u.ResultsUsing H&E staining alone, gross cT4a was most often pathologically underdiagnosed as pT3 (93.8%) and very rarely as pT4a, resulting in a large diagnostic discrepancy. Using EVG staining, 60.7% of the cT4a tumors were diagnosed as pT3b. The 10-year DFS and OS rates were similar for pT3a and pT3u patients. However, the 10-year DFS and OS rates of pT3b patients were significantly lower than those of pT3a patients (75.6% vs. 95.6%, p < 0.0001 and 58.4% vs. 70.6%, p = 0.0024, respectively) but did not differ from those of pT4a patients (70.6%, p = 0.5799 and 52.0%, p = 0.1116, respectively). Multivariate analysis revealed that the ELI was the strongest independent risk factor for recurrence and CRC-specific death (p < 0.0001).ConclusionsA better understanding of the ELI allows us to reconsider the diagnostic discrepancy of serosal invasion, i.e., pT3b should be considered pT4a. The ELI-based subclassification of pT3 is expected to be incorporated into the TNM staging system in the future. The ELI is a notable prognostic indicator in patients with pStage II CRC.

Assessing para-aortic nodal status in high-grade endometrial cancer patients with negative pelvic sentinel lymph node biopsy.

To determine the accuracy of pelvic sentinel lymph node biopsy (SLN) in detecting positive para-aortic (PA) lymph nodes in high-grade uterine cancer, and to determine the recurrence rate in patients with high-grade uterine cancers who did not receive adjuvant chemotherapy based on negative pelvic SLNs. This was a retrospective cohort study of patients with newly diagnosed, high-grade endometrial cancer who underwent surgery, including pelvic SLNs with or without PA node dissection, at a tertiary care institution between 2015 and 2020. Baseline demographics, surgical management, pathology data, and outcomes were analyzed using descriptive statistics, and survival analysis. Postoperative histology of the 110 patients meeting inclusion criteria was 45.5% grade 3 endometrioid, 36.4% serous, 10.9% clear cell, and 7.3% carcinosarcoma. On final pathology, 63.7% were stage 1, and 23.6% were stage 3C with positive nodes. A total of 63 patients (57.3%) had a PA lymph node dissection (56 bilateral, 7 unilateral) in addition to the pelvic SLN. Among this group, 5.8% (95% confidence interval 1.2%-16.0%) had a positive PA node despite a negative pelvic SLN. Among those with a negative pelvic SLN and no adjuvant chemotherapy (n = 75), the rate of distant recurrence was 14.7%, and 3-year recurrence-free survival was 71.9%. The rate of isolated PA node metastasis in high-grade endometrial cancers despite a negative pelvic SLN may be significantly higher than the accepted rate of isolated PA node metastasis in low-grade endometrial cancer. This supports adjuvant treatment decisions continuing to incorporate primary tumor pathology and molecular classification.

Development of predictive models for pathological response status in breast cancer after neoadjuvant therapy based on peripheral blood inflammatory indexes

BackgroundAchieving a pathological complete response (pCR) after neoadjuvant therapy (NAT) is considered to be a critical factor for a favourable prognosis in breast cancer. However, discordant pathological complete response (DpCR), characterised by isolated responses in the breast or axillary, represents an intermediate pathological response category between no response and complete response. This study aims to investigate predictive factors and develop models based on peripheral blood inflammatory indexes to more accurately predict NAT outcomes.MethodA total of 789 eligible patients were enrolled in this retrospective study. The patients were randomized into training and validation cohort according to a 7:3 ratio. Lasso and uni/multivariable logistic regression analysis were applied to identify the predictor variables. Two Nomograms combining clinico-pathologic features and peripheral blood inflammatory indexes were developed.ResultMolecular Subtype, HALP, P53, and FAR were used to construct the predictive models for traditional non pCR (T-NpCR) and total-pCR (TpCR). The T-NpCR group was divided into DpCR and non pCR (NpCR) subgroups to construct a new model to more accurately predict NAT outcomes. cN, HALP, FAR, Molecular Subtype, and RMC were used to construct the predictive models for NpCR and DpCR. The receiver operating characteristic (ROC) curves indicate that the model exhibits robust predictive capacity. Clinical Impact Curves (CIC) and Decision Curve Analysis (DCA) indicate that the models present a superior clinical utility.ConclusionHALP and FAR were identified as peripheral blood inflammatory index predictors for accurately predicting NAT outcomes.

Changes in WHO classification of adrenal tumors and new ideas for multi-dimensional diagnosis and treatment

In 2022, WHO updated the classification and concept of adrenal cortical and medullary tumors. In terms of adrenal cortical tumors, the WHO classification further standardizes the nomenclature of nodular adrenal cortical disease and refines the pathological classification of primary aldosteronism. In terms of adrenal medullary tumors, the WHO classification unifies the concepts of pheochromocytoma and paraganglioma, and reclassifies various concepts, including paraganglioma-like neuroendocrine tumors. The new standards not only cover the clinical manifestations of the disease, but also include other multiple aspects such as the histological origin of the disease, immunohistochemical manifestations, physiological mechanisms of the disease, hereditary susceptibility and prognostic factors. This article intends to explore how to improve the diagnostic and therapeutic level of adrenal tumors.

Automatic maxillary sinus segmentation and pathology classification on cone-beam computed tomographic images using deep learning

BackgroundMaxillofacial complex automated segmentation could alternative traditional segmentation methods to increase the effectiveness of virtual workloads. The use of DL systems in the detection of maxillary sinus and pathologies will both facilitate the work of physicians and be a support mechanism before the planned surgeries.ObjectiveThe aim was to use a modified You Only Look Oncev5x (YOLOv5x) architecture with transfer learning capabilities to segment both maxillary sinuses and maxillary sinus diseases on Cone-Beam Computed Tomographic (CBCT) images.MethodsData set consists of 307 anonymised CBCT images of patients (173 women and 134 males) obtained from the radiology archive of the Department of Oral and Maxillofacial Radiology. Bilateral maxillary sinuses CBCT scans were used to identify mucous retention cysts (MRC), mucosal thickenings (MT), total and partial opacifications, and healthy maxillary sinuses without any radiological features.ResultsRecall, precision and F1 score values for total maxillary sinus segmentation were 1, 0.985 and 0.992, respectively; 1, 0.931 and 0.964 for healthy maxillary sinus segmentation; 0.858, 0.923 and 0.889 for MT segmentation; 0.977, 0.877 and 0.924 for MRC segmentation; 1, 0.942 and 0.970 for sinusitis segmentation.ConclusionThis study demonstrates that maxillary sinuses can be segmented, and maxillary sinus diseases can be accurately detected using the AI model.

A novel embedded kernel CNN-PCFF algorithm for breast cancer pathological image classification

Early screening of breast cancer through image recognition technology can significantly increase the survival rate of patients. Therefore, breast cancer pathological image is of great significance for medical diagnosis and clinical research. In recent years, numerous deep learning models have been applied to breast cancer image classification, with deep CNN being a typical representative. Due to the use of multi-depth small convolutional kernels in mainstream CNN architectures such as VGG and Inception, the obtained image features often have high dimensionality. Although high dimensionality can bring more fine-grained features, it also increases the computational complexity of subsequent classifiers and may even lead to the curse of dimensionality and overfitting. To address these issues, a novel embedded kernel CNN principal component feature fusion (CNN-PCFF) algorithm is proposed. The constructed kernel function is embedded in the principal component analysis to form the multi-kernel principal component. Multi-kernel principal component analysis is used to fuse the high dimensional features obtained from the convolution base into some representative comprehensive variables, which are called kernel principal components, so as to achieve the purpose of dimensionality reduction. Any type of classifier can be added based on multi-kernel principal components. Through experimental analysis on two public breast cancer image datasets, the results show that the proposed algorithm can improve the performance of the current mainstream CNN architecture and subsequent classifiers. Therefore, the proposed algorithm in this paper is an effective tool for the classification of breast cancer pathological images.

Vision transformer introduces a new vitality to the classification of renal pathology

Recent advancements in computer vision within the field of artificial intelligence (AI) have made significant inroads into the medical domain. However, the application of AI for classifying renal pathology remains challenging due to the subtle variations in multiple renal pathological classifications. Vision Transformers (ViT), an adaptation of the Transformer model for image recognition, have demonstrated superior capabilities in capturing global features and providing greater explainability. In our study, we developed a ViT model using a diverse set of stained renal histopathology images to evaluate its effectiveness in classifying renal pathology. A total of 1861 whole slide images (WSI) stained with HE, MASSON, PAS, and PASM were collected from 635 patients. Renal tissue images were then extracted, tiled, and categorized into 14 classes on the basis of renal pathology. We employed the classic ViT model from the Timm library, utilizing images sized 384 × 384 pixels with 16 × 16 pixel patches, to train the classification model. A comparative analysis was conducted to evaluate the performance of the ViT model against traditional convolutional neural network (CNN) models. The results indicated that the ViT model demonstrated superior recognition ability (accuracy: 0.96–0.99). Furthermore, we visualized the identification process of the ViT models to investigate potentially significant pathological ultrastructures. Our study demonstrated that ViT models outperformed CNN models in accurately classifying renal pathology. Additionally, ViT models are able to focus on specific, significant structures within renal histopathology, which could be crucial for identifying novel and meaningful pathological features in the diagnosis and treatment of renal disease.

Significantly improving zero-shot X-ray pathology classification via fine-tuning pre-trained image-text encoders

Deep neural networks are increasingly used in medical imaging for tasks such as pathological classification, but they face challenges due to the scarcity of high-quality, expert-labeled training data. Recent efforts have utilized pre-trained contrastive image-text models like CLIP, adapting them for medical use by fine-tuning the model with chest X-ray images and corresponding reports for zero-shot pathology classification, thus eliminating the need for pathology-specific annotations. However, most studies continue to use the same contrastive learning objectives as in the general domain, overlooking the multi-labeled nature of medical image-report pairs. In this paper, we propose a new fine-tuning strategy that includes positive-pair loss relaxation and random sentence sampling. We aim to improve the performance of zero-shot pathology classification without relying on external knowledge. Our method can be applied to any pre-trained contrastive image-text encoder and easily transferred to out-of-domain datasets without further training, as it does not use external data. Our approach consistently improves overall zero-shot pathology classification across four chest X-ray datasets and three pre-trained models, with an average macro AUROC increase of 4.3%. Additionally, our method outperforms the state-of-the-art and marginally surpasses board-certified radiologists in zero-shot classification for the five competition pathologies in the CheXpert dataset.