Deep Learning Based Automated Orthotopic Lung Tumor Segmentation in Whole-Body Mouse CT-Scans

Abstract

Lung cancer is the leading cause of cancer related deaths worldwide. The development of orthotopic mouse models of lung cancer, which recapitulates the disease more realistically compared to the widely used subcutaneous tumor models, is expected to critically aid the development of novel therapies to battle lung cancer or related comorbidities such as cachexia. However, follow-up of tumor take, tumor growth and detection of therapeutic effects is difficult, time consuming and requires a vast number of animals in orthotopic models. Here, we describe a solution for the fully automatic segmentation and quantification of orthotopic lung tumor volume and mass in whole-body mouse computed tomography (CT) scans. The goal is to drastically enhance the efficiency of the research process by replacing time-consuming manual procedures with fast, automated ones. A deep learning algorithm was trained on 60 unique manually delineated lung tumors and evaluated by four-fold cross validation. Quantitative performance metrics demonstrated high accuracy and robustness of the deep learning algorithm for automated tumor volume analyses (mean dice similarity coefficient of 0.80), and superior processing time (69 times faster) compared to manual segmentation. Moreover, manual delineations of the tumor volume by three independent annotators was sensitive to bias in human interpretation while the algorithm was less vulnerable to bias. In addition, we showed that besides longitudinal quantification of tumor development, the deep learning algorithm can also be used in parallel with the previously published method for muscle mass quantification and to optimize the experimental design reducing the number of animals needed in preclinical studies. In conclusion, we implemented a method for fast and highly accurate tumor quantification with minimal operator involvement in data analysis. This deep learning algorithm provides a helpful tool for the noninvasive detection and analysis of tumor take, tumor growth and therapeutic effects in mouse orthotopic lung cancer models.

Keywords: artificial intelligence; deep learning; lung cancer; lung tumor segmentation; µCBCT.

Figure 1

Quantitative performance metrics of deep learning algorithm. (A) Box plots showing the distribution of the tumor volumes in mm³ per cross-validation batch (n = 15), determined by manual and automatic segmentation. (B–D) Box plots of (B) the dice similarity coefficient, (C) 95th percentile Hausdorff Distance (in mm), and (D) the center of mass displacement (in mm), calculated for each of the batches.

Figure 2

Qualitative evaluation of the algorithm performance. (A) Scatterplot of the DSC versus tumor volume (n = 60). The mean DSC is indicated with a horizontal red line. The ±1SD and ±2SD are indicated with blue dotted lines. (B) Axial µCBCT images showing (from left to the right) an unsegmented scan, the manual tumor segmentation, the automatic tumor segmentation, and the overlay of the manual and automatic segmentations. The first row shows an illustrative case with a high overlap, the second row shows a case with minimal overlap, and the third row depicts a case with medium overlap. The corresponding DSCs of each row in (B) are indicated in the scatterplot of (A) by the green, red and yellow markers.

Figure 3

Longitudinal follow-up of tumor volume. A series of axial slices of µCBCT images showing the manual segmentation (blue) and automatic segmentation (red) of the lung tumor in a single animal versus time. The µCBCT images represent one mouse at seven different time points. The quantified manual (blue) and automatically (red) segmented volumes are presented over time in the line graph.

Figure 4

Longitudinal follow-up of tumor volume and muscle mass. (A) A 3D image of a whole-body mouse µCBCT scan indicating the region of the axial slices selected for (B). (B) A series of axial slices of µCBCT images illustrating the automatic lung tumor segmentations (light blue) and lower hind limb muscle complex segmentations (orange). The µCBCT images represent one mouse at seven different time points. The quantified tumor volumes (light blue) and muscle masses (orange) are presented over time in the line graph.