Prognostic value of tumor volume for patients with advanced lung cancer treated with chemotherapy.

ABSTRACT

BACKGROUND AND OBJECTIVE: We aim to develop a reference system utilizing computed tomography to calculate changes in tumor volume of lung cancer patients after chemotherapy to assist physicians in clinical treatment and evaluation. METHODS: Image processing techniques were used to analyze the computed tomography of lung cancer, locate the tumor, and calculate the tumor volume. The medical indicator was then evaluated and analyzed. We examined the correlation between reduced tumor volume and survival duration of 88 patients after chemotherapy at Tri-Service General Hospital, Taiwan. The innovative survival prediction index was obtained by four statistical methods: receiver operating characteristic curve, Youden index, Kaplan-Meier method, and log rank test. RESULTS: From the image processing techniques, tumor volume from each patient were obtained within an average of 7.25 seconds. The proposed method was shown to achieve rapid positioning of lung tumors and volume reconstruction with an estimation error of 1.92% when calibrated with an irregularly shaped stone. In medical indicator evaluation and analysis, the area below the receiver operating characteristic curve is greater than 0.8, indicating good predictability of the medical index used herein. The Youden index spotted the best cut-off point of volume, and the correlation between the volume's cut-off point and survival time was confirmed again by Kaplan-Meier and log rank test. The p-values were all less than 0.05, presenting a high degree of correlation between the two, indicating that this medical indicator is highly reliable. CONCLUSIONS: The proposed techniques can automatically find the location of tumors in the lung, reconstruct the volume, and calculate changes in volume before and after treatment, thus obtaining an innovative survival prediction index. This will help facilitate early and accurate predictions of disease outcomes during the course of therapy, and categorize patient stratification into risk groups for more efficient therapies.