Diagnosis Of Lymph Node Disease Research Articles

Segmentation of Lymph Nodes in Ultrasound Images Using U-Net Convolutional Neural Networks and Gabor-Based Anisotropic Diffusion

PurposeThe automated segmentation of lymph nodes (LNs) in ultrasound images is challenging, largely because of speckle noise and echogenic hila. This paper proposes a fully automatic and accurate method for LN segmentation in ultrasound that overcomes these issues.MethodsThe proposed segmentation method integrates diffusion-based despeckling, U-Net convolutional neural networks and morphological operations. First, the speckle noise is suppressed and the lymph node edges are enhanced using Gabor-based anisotropic diffusion (GAD). Then, a modified U-Net model is used to segment the LNs excluding any echogenic hila. Finally, morphological operations are undertaken to segment the entire LNs by filling in any regions occupied by echogenic hila.ResultsA total of 531 lymph nodes from 526 patients were segmented using the proposed method. Its segmentation performance was evaluated in terms of its accuracy, sensitivity, specificity, Jaccard similarity and Dice coefficient, for which it achieved values of 0.934, 0.939, 0.937, 0.763 and 0.865, respectively.ConclusionThe proposed method automatically and accurately segments LNs in ultrasound images, enhancing the prospects of being able to undertake artificial intelligence (AI)-based diagnosis of lymph node diseases.

Staging Strategies for Lung Cancer

LUNG CANCER HAS BECOME PERHAPS THE MOST IMPORtant cancer in the United States. The incidence has continued to increase, and lung cancer is now the leading cause of cancer death. However, most patients with lung cancer have advanced-stage disease when they first present and are not amenable to curative treatment. For the remainder, a careful search for metastatic or locally invasive tumors must be made prior to potentially curative lung resection. The lymph node drainage system for the lungs lies primarily in the mediastinum, and because it is a compact space enclosed in a rigid, bony cage, finding small lymph nodes that contain cancer is challenging. Computed tomographic (CT) evaluation of the chest reliably identifies and characterizes lung tumors but is less useful for evaluating the mediastinum for malignant disease. Mediastinal evaluation by CT is only 51% sensitive (95% confidence interval [CI], 47%-54%) and 85% specific (95% CI, 84%-88%) for establishing the diagnosis of lymph node disease in lung cancer. Consequently, mediastinoscopy is a required intermediary step between an otherwise negative metastatic evaluation for lung cancer and definitive pulmonary resection. Although complications are infrequent, mediastinoscopy is an operative procedure requiring general anesthesia and can substantially add to the overall morbidity and cost of lung cancer treatment. Positron emission tomographic (PET) scanning with the glucose analog fluorodeoxyglucose offered great promise for imaging tumors. Because cancers have higher-thannormal glucose uptake coupled with lower phosphorylation rates, fluorodeoxyglucose preferentially accumulates in malignant cells. Early studies showed great promise for this noninvasive radionucleide scanning technique to identify malignant cells too small to detect by CT. Nevertheless, inflammatory nodes resulting in false-positive results and lymph nodes too small for the resolution of current PET scanners result in a disappointingly low sensitivity of 74% (95% CI, 69%-79%) and specificity of 85% (95% CI, 82%-88%). Current clinical practice guidelines recommend that patients with lung cancer undergo CT scanning with the intent of identifying lymph node metastases and, if this imaging evaluation is unrevealing, that they next undergo PET scanning. With this strategy, some patients will still have metastatic disease in the mediastinum after noninvasive imaging. Currently, these patients undergo mediastinoscopy prior to thoracotomy in an effort to reduce the rate of unnecessary chest explorations that are potentially aborted because of previously unrecognized unresectable or metastatic lung cancer. In recent years, advances in endoluminal ultrasound (EUS) technologies facilitated high-resolution imaging of structures adjacent to the esophagus and bronchus. These devices enable biopsy of suspicious mediastinal lesions with the promise of replacing or eliminating PET imaging and mediastinoscopy in the evaluation of patients with lung cancer. Indeed, the combination of EUS and endobronchial ultrasound (EBUS) with fine-needle aspiration in nonrandomized trials demonstrated a 93% sensitivity (95% CI, 81%-99%) and 97% specificity (95% CI, 91%-99%) for establishing the presence of mediastinal disease in lung cancer patients. These impressive results suggest that the combination of EUS and EBUS could replace mediastinoscopy as well as avoid thoracotomies that proved unnecessary because of the incidental finding of advanced-stage disease. In this issue of JAMA, Annema and colleagues report the results of a randomized trial comparing mediastinoscopy and endosonography for mediastinal nodal staging of lung cancer. This study highlights several important issues regarding the importance of properly staging lung cancer preoperatively to avoid surgical intervention in a group of patients who have no proven benefit from resection and in fact would have a reduction in their quality of life from an unnecessary thoracotomy. This randomized study compares the use of endoscopic ultrasound via either esophagoscopy, bronchoscopy, or both, followed by surgical staging if no nodal metastases were found, to the past gold standard of surgical staging either by mediastinoscopy or parasternal biopsy. This welldesigned study attempts to answer the question of how this

Mantle Cell Lymphoma Genotypes Identified with CGH to BAC Microarrays Define Clinical Subgroups of Disease and Strongly Predict Patient Outcome.

Mantle cell lymphoma (MCL) is a rare, aggressive subtype of B-cell non-Hodgkin lymphoma (B-NHL), characterized by t(11;14)(q13;q32). To identify recurrent genomic events involved in the pathogenesis of this malignancy, we used genome-wide comparative genomic hybridization (CGH) to 1.4 Mb. resolution BAC microarrays in 68 patients and 9 derived MCL cell lines; the genome profiles were compared to those from other B-NHL subtypes. Array-CGH defined a MCL genomic signature different from other B-cell lymphomas, including deletions of chromosomes 1p21.2-p12.3, 11q22.3 at ATM gene locus, 13q14.2 and 13q34, with coincident 10p12-BMI locus amplification and 10p14 deletion, along with a previously unidentified loss within 9q21-q22. Specific genomic alterations were associated with different subgroups of disease. Notably, 11 patients with non-nodal, leukemic MCL, defined on the basis of the absence at diagnosis of lymph node disease, showed a different genomic profile to nodal cases, including deletion of 8p21.2 at TRAIL receptor gene cluster (55 vs. 19%;p=0,01) and gain of 8q24.1 at MYC locus (46 vs. 14%;p=0,015). Additionally, leukemic MCL exhibited frequent IgVH mutation (64 vs. 21%;p=0,009) with preferential VH-39 use (36 vs. 4%;p=0,005), and followed a more indolent course, being 7 of 11 patients (64%) survivors for more than 3 years whereas only 15 of 55 patients (27%) in the nodal group experienced this long survival (median survial time, 42 vs. 18 months; p=0,02). The median overall survival (OS) for the patients in the series was 39 months (range, 0–115). Blastoid variant of MCL, increased number of genomic gains, and deletions of P16/INK4a and P53 genes correlated with poorer outcomes, whilst 1p21 loss was associated with prolonged survival (median OS, 70 vs. 31 months; p=0,02). In multivariate analysis, deletion of chromosome 9q21-q22, which has not been previously reported as a common change in MCL, was the strongest predictor for inferior survival (HR:6; CI:2,3–15,7). Last, the genotypes of 14 patients with MCL who were alive for more than 5 years since diagnosis showed deletion of 1p21 in seven cases (50%), but none exhibited deletions of 9q21-q22, 9p21.3-P16/INK4a or 17p13.1-P53. Our study illustrates the utility of array-CGH in MCL sub-classification and highlights the array genomic profile as the strongest indicator for predicting clinical outcome. The screening of this reduced BAC clone set either by FISH or custom-made array CGH devices could be reliably applied to the clinical diagnostics of MCL.

Wide needle aspiration cytology in the diagnosis of lymphadenopathy in Zambia.

To study the value of wide needle (19 gauge) aspiration cytology in the diagnosis of lymph node disease in Zambia in the absence of a trained cytologist. Patients (n = 304) referred for surgical biopsy of an enlarged peripheral lymph node were studied prospectively. Surgical biopsy was routinely preceded by 19 gauge needle aspiration of the same node; aspirates were stained by haematoxylin and eosin and Ziehl Neelsen stains. Of 232 aspirates, 182 contained sufficient material for cytological characterisation. Tuberculosis was diagnosed or suspected in 122 of 126 aspirates with histologically confirmed tuberculous lymphadenitis; reactive follicular hyperplasia in 31 of 38 patients with primary HIV lymphadenopathy; malignancy in all five patients with malignant nodes; and Kaposi's disease in four of nine patients with this. Tuberculous lymphadenitis was falsely suspected in four patients, as was reactive follicular hyperplasia in four, and Kaposi's disease in four. Wide needle aspiration cytology is useful in the diagnosis of lymphadenopathy in Central Africa, with the exception of lymphadenopathic Kaposi's disease.

An evaluation of the diagnostic accuracy of Pathfinder

We present an evaluation of the diagnostic accuracy of Pathfinder, an expert system that assists pathologists with the diagnosis of lymph node diseases. We evaluate two versions of the system using both informal and decision-theoretic metrics of performance. In one version of Pathfinder, we assume incorrectly that all observations are conditionally independent. In the other version, we use a belief network to represent accurately the probabilistic dependencies among the observations. In both versions, we make the assumption—reasonable for this domain—that diseases are mutually exclusive and exhaustive. The results of the study show that (1) it is cost effective to represent probabilistic dependencies among observations in the lymph node domain, and (2) the diagnostic accuracy of the more complex version of Pathfinder is at least as good as that of the Pathfinder expert. In addition, the study illustrates how informal and decision-theoretic metrics for performance complement one another.

Toward Normative Expert Systems: Part I The Pathfinder Project

Pathfinder is an expert system that assists surgical pathologists with the diagnosis of lymph-node diseases. The program is one of a growing number of normative expert systems that use probability and decision theory to acquire, represent, manipulate, and explain uncertain medical knowledge. In this article, we describe Pathfinder and our research in uncertain-reasoning paradigms that was stimulated by the development of the program. We discuss limitations with early decision-theoretic methods for reasoning under uncertainty and our initial attempts to use non-decision-theoretic methods. Then, we describe experimental and theoretical results that directed us to return to reasoning methods based in probability and decision theory.

Toward Normative Expert Systems: Part II Probability-Based Representations for Efficient Knowledge Acquisition and Inference

We address practical issues concerning the construction and use of decision-theoretic or normative expert systems for diagnosis. In particular, we examine Pathfinder, a normative expert system that assists surgical pathologists with the diagnosis of lymph-node diseases, and discuss the representation of dependencies among pieces of evidence within this system. We describe the belief network, a graphical representation of probabilistic dependencies. We see how Pathfinder uses a belief network to construct differential diagnosis efficiently, even when there are dependencies among pieces of evidence. In addition, we introduce an extension of the belief-network representation called a similarity network, a tool for constructing large and complex belief networks. The representation allows a user to construct independent belief networks for subsets of a given domain. A valid belief network for the entire domain can then be constructed from the individual belief networks. We also introduce the partition, a graphical representation that facilitates the assessment of probabilities associated with a belief network. We show that the similarity-network and partition representations made practical the construction of Pathfinder.

Automated Cytochemistry in Non-Hodgkin’s Lymphoma: A New Method for Determination of Cells from Lymph Node Biopsy

Cell suspension prepared from the lymph node biopsy of patients with non-Hodgkin's lymphoma (NHL), metastatic carcinoma (MC) and non-neoplastic lymphadenopathies (NL) were analyzed by the Hemalog D, automated differential counter. The preparation of lymph node cells is described first in this study. The results show that the percentage of large cells (diameter greater than 13.5 micron) stained negatively with peroxidase (LUC, large unstained cells) was remarkably increased in patients with NHL (mean +/- SEM = 18.6 +/- 3.1%) and was particularly increased in the subgroup, diffuse histiocytic type (31.1 +/- 5.3%). Patients with MC had a raised percentage of nonspecific esterase-positive cells (9.2 +/- 1.4%) compared to patients in the NHL and NL groups. Patients in the NL group had low percentages of both LUC (5.3 +/- 0.7%) and nonspecific esterase-positive cells (1.8 +/- 0.2%). Quantitation of cells in the lymph node by using the Hemalog D may assist in the diagnosis of lymph node diseases.