756 SONAZOID CONTRAST ENHANCED ULTRASONOGRAPHY COULD BE USED FOR MONITORING OF SORAFENIB'S EARLY RESPONSE AND FOR FOLLOW UP OF ADVANCED HEPATOCELLULAR CARCINOMA?

Abstract

for early diagnosis and improved survival. Vibrational spectroscopy methods like Raman spectroscopy have shown their potential to identify and quantify various biomolecules present in body fluids. In this pilot study, we investigated whether Raman microspectroscopy can differentiate sera from cirrhotic patients with and without HCC. Methods: Serum samples were collected from 2 sets of patients: cirrhotic patients without HCC (n =39) and with HCC (n =34). Micro-Raman spectra were acquired from dried drops of serum in the point-mode using a LabRam microspectrometer (Horiba Scientific, France) equipped with 100x objective and a 785 nm laser. For each sample, 5 replicate spectra were recorded, each during 40 s in the spectral range 600–1800 cm. Spectra were quality-tested and preprocessed (smoothing, baseline subtraction, vectorial normalisation). Then, a principal component (PC) analysis was performed and the support vector machine (SVM) method was applied using the most discriminant PCs to classify the spectra into 2 classes of cirrhotic patients with and without HCC. Results: Micro-Raman spectra from 4 patients did not meet the quality test and were excluded. SVM was performed on a training set (25 patients with HCC and 21 patients without HCC) to construct the model, a validation set (9 patients of each group) to choose the model with the best generalized predictive value, and an external test set (5 and 4 patients of each group) to confirm the predictive accuracy of the model for the diagnosis of HCC. The model presenting the best training and validation accuracy was obtained with 17 PCs. The results of this model on the three datasets are shown below: