Michael Hareskov Larsen

Efficacy of an Artificial Neural Network–Based Approach to Endoscopic Ultrasound Elastography in Diagnosis of Focal Pancreatic Masses

BACKGROUND & AIMS By using strain assessment, real-time endoscopic ultrasound (EUS) elastography provides additional information about a lesions characteristics in the pancreas. We assessed the accuracy of real-time EUS elastography in focal pancreatic lesions using computer-aided diagnosis by artificial neural network analysis. METHODS We performed a prospective, blinded, multicentric study at of 258 patients (774 recordings from EUS elastography) who were diagnosed with chronic pancreatitis (n = 47) or pancreatic adenocarcinoma (n = 211) from 13 tertiary academic medical centers in Europe (the European EUS Elastography Multicentric Study Group). We used postprocessing software analysis to compute individual frames of elastography movies recorded by retrieving hue histogram data from a dynamic sequence of EUS elastography into a numeric matrix. The data then were analyzed in an extended neural network analysis, to automatically differentiate benign from malignant patterns. RESULTS The neural computing approach had 91.14% training accuracy (95% confidence interval [CI], 89.87%-92.42%) and 84.27% testing accuracy (95% CI, 83.09%-85.44%). These results were obtained using the 10-fold cross-validation technique. The statistical analysis of the classification process showed a sensitivity of 87.59%, a specificity of 82.94%, a positive predictive value of 96.25%, and a negative predictive value of 57.22%. Moreover, the corresponding area under the receiver operating characteristic curve was 0.94 (95% CI, 0.91%-0.97%), which was significantly higher than the values obtained by simple mean hue histogram analysis, for which the area under the receiver operating characteristic was 0.85. CONCLUSIONS Use of the artificial intelligence methodology via artificial neural networks supports the medical decision process, providing fast and accurate diagnoses.

Endoscopic ultrasound, endoscopic sonoelastography, and strain ratio evaluation of lymph nodes with histology as gold standard

BACKGROUND AND STUDY AIMS Accurate lymph node staging is essential for the selection of an optimal treatment in patients with upper gastrointestinal cancer. Endoscopic ultrasound (EUS) and fine-needle aspiration (FNA) are considered to be the most accurate method for locoregional staging. Endoscopic sonoelastography (ESE) assesses the elasticity of lymph nodes and has been used to differentiate lymph nodes with promising results. The aim of this study was to evaluate the use of EUS, EUS - FNA, ESE, and ESE-strain ratio using histology as the gold standard. PATIENTS AND METHODS Patients with upper gastrointestinal cancer who were referred for EUS examination were enrolled if surgical treatment was planned and the patient had a lymph node that was accessible for EUS - FNA and EUS-guided fine-needle marking (FNM). The lymph node was classified using EUS, ESE, and ESE-strain ratio. Finally, EUS - FNA and EUS - FNM were performed. The marked lymph node was isolated during surgery for histological examination. RESULTS The marked lymph node was isolated for separate histological examination in 56 patients, of whom 22 (39 %) had malignant lymph nodes and 34 (61 %) had benign lymph nodes. There were no complications of EUS - FNM. The sensitivity of EUS for differentiation between malignant and benign lymph nodes was 86 % compared with 55 % - 59 % for the different ESE modalities. The specificity of EUS was 71 % compared with 82 % - 85 % using ESE modalities. CONCLUSION The use of the EUS - FNM technique enabled the identification of a specific lymph node and thereby the use of histology as gold standard. ESE and ESE-strain ratio were no better than standard EUS in differentiating between malignant and benign lymph nodes in patients with resectable upper gastrointestinal cancer.

Intra- and Interobserver Agreement of Endoscopic Sonoelastography in the Evaluation of Lymph Nodes

PURPOSE There is a lack of studies on the intra- and interobserver variability of endoscopic ultrasonography (EUS) and especially endoscopic sonoelastography (ES). The aim of this study was to evaluate the intra- and interobserver agreement of EUS and ES during the evaluation of a specific lymph node in patients with upper gastrointestinal malignancies (UGIM). The ES strain ratio was used to differentiate between benign and malignant lymph nodes and the interobserver agreement was evaluated. MATERIALS AND METHODS This study prospectively enrolled 62 patients with UGIM. EUS and ES were performed by two observers in randomized order of a specific lymph node using a linear Pentax echoendoscope and the Hitachi EUB-8500 US unit. The lymph node was classified as malignant or benign on the basis of EUS, ES, an ES scoring system, and 5 repeated strain ratio measurements. The intraobserver variation was evaluated by each observer performing 20 strain ratio measurements of the same lymph node. Finally, EUS fine needle aspiration was performed. The kappa statistic was used to test for interobserver variability. RESULTS One patient was excluded due to inadequate ES imaging. Thus, 61 patients were included in the analysis of the interobserver agreement. Using EUS, ES, and an ES scoring system, the kappa values were 0.80, 0.58, and 0.35, respectively. An ES strain ratio of 3.81 was defined as the cut-off value between benign and malignant lymph nodes using cytology as the gold standard (n = 55). Using this modality, a kappa value of 0.59 was obtained. A t-test comparison of the measured strain ratios for the two observers found no significant differences. CONCLUSION ES and ES strain ratio evaluation of lymph nodes were feasible and may be reproduced with good interobserver agreement in a blinded clinical setup. The predefined ES scoring system provided only poor interobserver agreement. Image selection should be part of the intra- and interobserver evaluation. ES strain ratio seemed promising but larger studies are needed to evaluate this new feature.

Laparoscopic ultrasound-guided biopsy in upper gastrointestinal tract cancer patients

BackgroundNoninvasive pretherapeutic staging may be supplemented with laparoscopy and laparoscopic ultrasonography (LUS) in order to detect minute liver metastases, carcinosis or other signs of nonresectable or disseminated disease in patients with upper gastrointestinal tract cancer (UGIC). The aim of this study was to evaluate the use, potential clinical gain, and safety profile of LUS-guided biopsy in patients with UGIC.MethodsA prospective consecutive study on LUS-guided biopsy in patients referred with UGIC between May 2007 and May 2008 was carried out. Previous noninvasive imaging methods had found no signs of disseminated disease. Laparoscopic or LUS-guided biopsies were only performed if a malignant result would change patient management.ResultsTwo hundred and nine patients entered the study and, based on predefined biopsy indications, laparoscopy and LUS-guided biopsies changed patient management in a total of 27.3% (54/198) of the patients with a final malignant diagnosis. There were no complications. Liver and pancreas were the main target areas for LUS-guided biopsies, and more than half of the biopsies (55%) were taken from the primary tumor where other modalities had failed to obtain proof of malignancy. Twenty-six percent of biopsies were taken from a suspected metastatic lesion not seen before, whereas 19% were taken from previously suspected metastases where other imaging modalities had failed to obtain proof of malignancy.ConclusionLUS-guided biopsy is a safe procedure which in combination with laparoscopic biopsies had an impact on patient management in one-quarter of UGIC patients.

Endoscopic ultrasound-guided fine-needle marking of lymph nodes

BACKGROUND AND STUDY AIMS: No previous studies have evaluated the ability of endoscopic ultrasonography to describe the anatomic location of lymph nodes on the basis of a node-to-node comparison. The aim of this study was to assess the feasibility and safety of a new endoscopic ultrasound (EUS)-guided fine-needle technique for marking lymph nodes. PATIENTS AND METHODS: Twenty-five patients with suspected or confirmed malignancies of the upper gastrointestinal tract were prospectively included. EUS-guided fine-needle marking (EUS-FNM) was performed with a silver pin with a diameter that allowed it to fit into a 19-gauge needle. The position of the pin was verified by EUS. End points were the ability to identify and isolate the marked lymph node during surgery and a comparison between the location of the pin as suggested by EUS and the actual location found in the resected specimen. RESULTS: Twenty-three lymph nodes were marked. Nineteen intended surgical isolations were performed. The lymph nodes were isolated in the resection specimens in 18 patients (95 %). In 2 out of 20 cases the pin was not localized by laparoscopic ultrasonography. In 89 % of the cases the marked lymph node was in the same location as described by EUS. One pin (5 %) was not retrieved. In three cases, a small hematoma was observed. There was no sign of long-term complications. CONCLUSION: EUS-FNM with a silver pin in lymph nodes is feasible and safe. EUS-FNM seems to be a suitable tool for evaluating lymph nodes on the basis of a node-to-node comparison.

Prospective evaluation of EUS-guided fine needle biopsy in pancreatic mass lesions

Background and study aim  Due to the scarcity of specific data on endoscopic ultrasound (EUS)-guided fine-needle biopsies (SharkCore) FNB in the evaluation of pancreatic lesions, we performed a prospective study of the diagnostic performance of EUS SharkCore FNB in patients with pancreatic lesions. The aim of this study was to evaluate the diagnostic accuracy. Patients and methods  Single-center prospective study of 41 consecutive patients referred for EUS-FNB from October 2015 to April 2016 at our center. EUS-FNB was obtained in a predefined setting regarding the procedure and pathological evaluation. Data regarding demographics, lesion, technical parameters, and diagnostic accuracy were obtained. Results  The study included 41 consecutive patients (22 males (54 %); median age 68 years). The average size of the lesions was 28 mm (median: 30 mm). A diagnostic specimen was identified in 40 (98 %) cases during microscopy with an average of 2.4 passes. The route was trans-duodenal in 20 cases (49 %). The histological diagnosis of the specimens was malignant in 29 cases (71 %), benign in 8 (20 %), suspicious in 2 (5 %), atypical in 1 (2 %) and in 1 (2 %) no material for microscopic evaluation was obtained. This led to a diagnostic accuracy of 93 %, sensitivity of 91 % and a specificity of 100 %. 2 cases (5 %) of self-limiting bleeding were observed. The diagnosis at follow up was malignant in 32 (78 %) of the patients. Conclusions  EUS-FNB of pancreatic mass lesions with the SharkCore needle produced specimens with a diagnostic accuracy of 93 %. The procedure was safe and easy to perform, and these data support the use of EUS-FNB in a routine setting.