Use of a convolutional neural network and quantitative ultrasound for diagnosis of fatty liver

Quantitative ultrasound (QUS) was used to classify rabbits that were induced to have liver disease by placing them on a fatty diet for a defined duration and/or periodically injecting them with CCl₄. The ground truth of the liver state was based on lipid liver percents estimated via the Folch assay and hydroxyproline concentration to quantify fibrosis. Rabbits were scanned ultrasonically in vivo using a SonixOne scanner and an L9-4/38 linear array. Liver fat percentage was classified based on the ultrasonic backscattered radiofrequency (RF) signals from the livers using either QUS or a 1-D convolutional neural network (CNN). Use of QUS parameters with linear regression and canonical correlation analysis demonstrated that the QUS parameters could differentiate between livers with lipid levels above or below 5%. However, the QUS parameters were not sensitive to fibrosis. The CNN was implemented by analyzing raw RF ultrasound signals without using separate reference data. The CNN outputs the classification of liver as either above or below a threshold of 5% fat level in the liver. The CNN outperformed the classification utilizing the QUS parameters combined with a support vector machine in differentiating between low and high lipid liver levels (i.e., accuracies of 74% versus 59% on the testing data). Therefore, although the CNN did not provide a physical interpretation of the tissue properties (e.g., attenuation of the medium or scatterer properties) the CNN had much higher accuracy in predicting fatty liver state and did not require an external reference scan.     

Late recurrences of pancreatic cancer in patients with long-term survival after pancreaticoduodenectomy

BackgroundPancreatic cancer remains a relevant clinical problem due to poor prognosis. Even after curative pancreaticoduodenectomy tumor recurrences occur in up to 80%. Risk factors for postoperative tumor recurrences have been identified before, but data on risk factors for tumor recurrences in patients with long-term-survival is scarce.MethodsIn this retrospective study consecutive long-term survival-patients (defined as at least 60 months survival) undergoing pancreaticoduodenectomy for pancreatic cancer from 2007–2014 were identified in the 2nd largest pancreatic surgery center in Germany. Clinical, pathohistological and laboratory values were analyzed to identify risk factors for tumor recurrence.ResultsThirty-four of one-hundred-sixty-seven patients were identified as long-term-survival-patients in the study period. Of those, 10 patients (29.4%) suffered from tumor recurrence. Lymph vessel invasion was identified as an independent risk factor (P=0.031, hazard ratio 13.127, 95% confidence interval: 1.270–135.698). Median postoperative time to tumor recurrence in long-term-survival-patients was 49 months. Overall survival after diagnosis of tumor recurrence was 33 months. 80% (N=8) of the patients were asymptomatic. Half of the patients (N=5) suffered from local disease, with 40% undergoing curative tumor resection. CA 19-9 levels were significantly elevated at 57 U/mL (normal <27 U/mL).ConclusionsTumor recurrence in long-term-survival-patients is typically asymptomatic. Especially long-term-survival-patients with lymph vessel invasion are more likely to develop tumor recurrence. Therefore, a structured follow-up program including CT-scans and CA 19-9 surveillance must be continued in all patients undergoing pancreaticoduodenectomy even in cases of long-term-survival.     

Inflammatory and regulatory CCL and CXCL chemokine and cytokine cellular responses in patients with patent Mansonella perstans filariasis

Mansonella perstans (Mp) filariasis is present in large populations in sub-Saharan Africa, and to what extent patent Mp infection modulates the expression of immunity in patients, notably their cellular cytokine and chemokine response profile, remains not well known. We studied the spontaneous and inducible cellular production of chemokines (C-X-C motif) ligand 9 (CXCL9) [monokine induced by interferon (IFN)-γ (MIG)], CXCL-10 [inducible protein (IP)-10], chemokine (C-C motif) ligand 24 (CCL24) (eotaxin-2), CCL22 [macrophage-derived chemokine (MDC)], CCL13 [monocyte chemotactic protein-4 (MCP-4)], CCL18 [pulmonary and activation-regulated chemokine (PARC)], CCL17 [thymus- and activation-regulated chemokine (TARC)] and interleukin (IL)-27 in mansonelliasis patients (Mp-PAT) and mansonelliasis-free controls (CTRL). Freshly isolated peripheral mononuclear blood cells (PBMC) were stimulated with helminth, protozoan and bacterial antigens and mitogen [phytohaemagglutinin (PHA)]. PBMC from Mp-PAT produced spontaneously (without antigen stimulation) significantly higher levels of eotaxin-2, IL-27, IL-8, MCP-4 and MDC than cells from CTRL, while IFN-γ-IP-10 was lower in Mp-PAT. Helminth antigens activated IL-27 and MCP-4 only in CTRL, while Ascaris antigen, Onchocerca antigen, Schistosoma antigen, Entamoeba antigen, Streptococcus antigen, Mycobacteria antigen and PHA stimulated MIG release in CTRL and Mp-PAT. Notably, Entamoeba antigen and PHA strongly depressed (P < 0·0001) eotaxin-2 (CCL24) production in both study groups. Multiple regression analyses disclosed in Mp-PAT and CTRL dissimilar cellular chemokine and cytokine production levels being higher in Mp-PAT for CCL24, IL-27, IL-8, MCP-4, MDC and PARC (for all P < 0·0001), at baseline (P < 0·0001), in response to Entamoeba histolytica strain HM1 antigen (EhAg) (P < 0·0001), Onchocerca volvulus adult worm-derived antigen (OvAg) (P = 0·005), PHA (P < 0·0001) and purified protein derivative (PPD) (P < 0·0001) stimulation. In Mp-PAT with hookworm co-infection, the cellular chemokine production of CXCL10 (IP-10) was diminished. In summary, the chemokine and cytokine responses in Mp-PAT were in general not depressed, PBMC from Mp-PAT produced spontaneously and selectively inducible inflammatory and regulatory chemokines and cytokines at higher levels than CTRL and such diverse and distinctive reactivity supports that patent M. perstans infection will not polarize innate and adaptive cellular immune responsiveness in patients.