A deep learning method for translating 3DCT to SPECT ventilation imaging: First comparison with 81m Kr-gas SPECT ventilation imaging

Abstract

Purpose: This study aimed to evaluate the accuracy of deep learning (DL)-based computed tomography (CT) ventilation imaging (CTVI).

Methods: A total of 71 cases that underwent single-photon emission CT ^81m Kr-gas ventilation (SPECT V) and CT imaging were included. Sixty cases were assigned to the training and validation sets, and the remaining 11 cases were assigned to the test set. To directly transform three-dimensional (3D) CT (free-breathing CT) images to SPECT V images, a DL-based model was implemented based on the U-Net architecture. The input and output data were 3DCT- and SPECT V-masked, respectively, except for whole-lung volumes. These data were rearranged in voxel size, registered rigidly, cropped, and normalized in preprocessing. In addition to a standard estimation method (i.e., without dropout during the estimation process), a Monte Carlo dropout (MCD) method (i.e., with dropout during the estimation process) was used to calculate prediction uncertainty. To evaluate the two models' (CTVI_{MCD U-Net} , CTVI_U-Net ) performance, we used fivefold cross-validation for the training and validation sets. To test the final model performances for both approaches, we applied the test set to each trained model and averaged the test prediction results from the five trained models to acquire the mean test result (bagging) for each approach. For the MCD method, the models were predicted repeatedly (sample size = 200), and the average and standard deviation (SD) maps were calculated in each voxel from the predicted results: The average maps were defined as test prediction results in each fold. As an evaluation index, the voxel-wise Spearman rank correlation coefficient (Spearman r_s ) and Dice similarity coefficient (DSC) were calculated. The DSC was calculated for three functional regions (high, moderate, and low) separated by an almost equal volume. The coefficient of variation was defined as prediction uncertainty, and these average values were calculated within three functional regions. The Wilcoxon signed-rank test was used to test for a significant difference between the two DL-based approaches.

Results: The average indexes with one SD (1SD) between CTVI_{MCD U-Net} and SPECT V were 0.76 ± 0.06, 0.69 ± 0.07, 0.51 ± 0.06, and 0.75 ± 0.04 for Spearman r_s , DSC_high , DSC_moderate , and DSC_low , respectively. The average indexes with 1SD between CTVI_U-Net and SPECT V were 0.72 ± 0.05, 0.66 ± 0.04, 0.48 ± 0.04, and 0.74 ± 0.06 for Spearman r_s , DSC_high , DSC_moderate , and DSC_low , respectively. These indexes between CTVI_{MCD U-Net} and CTVI_U-Net showed no significance difference (Spearman r_s , p = 0.175; DSC_high , p = 0.123; DSC_moderate , p = 0.278; DSC_low , p = 0.520). The average coefficient of variations with 1SD were 0.27 ± 0.00, 0.27 ± 0.01, and 0.36 ± 0.03 for the high-, moderate-, and low-functional regions, respectively, and the low-functional region showed a tendency to exhibit larger uncertainties than the others.

Conclusion: We evaluated DL-based framework for estimating lung-functional ventilation images only from CT images. The results indicated that the DL-based approach could potentially be used for lung-ventilation estimation.

Keywords: deep learning; functional imaging; radiotherapy.

FIGURE 1

Model architecture used within the study. This CNN model is based on the U‐Net architecture. The masked CT volume is input into the model, and the masked ventilation volume is output through the encoding and decoding paths. Numbers on the left side and top of the model represent the volume shape and the number of feature maps in a particular layer, respectively. CNN, convolutional neural network; Conv, convolutional layer; CT, computed tomography; Trans conv: transposed convolutional layer

FIGURE 2

Typical cases of coronal slices showing relatively good performance based on the Spearman r _s values between SPECT V and CTVIs of both models. The top row shows the SPECT V on the left, followed by the ventilation predictions of the CTVI_{MCD U‐Net} and the scatter plot between SPECT V and CTVI_{MCD U‐Net} within whole‐lung volume. The bottom row shows the masked CT on the left, followed by the ventilation predictions of the CTVI_U‐Net and the scatter plot between SPECT V and CTVI_U‐Net within whole‐lung volume. Orange arrows indicate the defect regions of SPECT V. The ventilation values for viewing were normalized to give the 99th percentile value of 1 and the 1st percentile value of 0. CT, computed tomography; CTVI, computed tomography ventilation imaging; MCD, Monte Carlo dropout; SPECT V, single‐photon emission computed tomography ventilation

FIGURE 3

Typical cases of coronal slices showing relatively poor performances based on the Spearman r _s values between SPECT V and CTVIs of both models. The top row shows the SPECT V on the left side, followed by the ventilation predictions of the CTVI_{MCD U‐Net} and the scatter plot between SPECT V and CTVI_{MCD U‐Net} within whole‐lung volume. The bottom row shows the masked CT on the left side, followed by the ventilation predictions of the CTVI_U‐Net and the scatter plot between SPECT V and CTVI_U‐Net within whole‐lung volume. Orange arrows indicate the defect regions of SPECT V. The ventilation values for viewing were normalized to give the 99th percentile value of 1 and the 1st percentile value of 0. CT, computed tomography; CTVI, computed tomography ventilation imaging, MCD, Monte Carlo dropout; SPECT V, single‐photon emission computed tomography ventilation

FIGURE 4

Boxplot and strip plot showing the bagged prediction performance for the 11 test cases for each of the 5 CNN models (MCD U‐Net and U‐Net) trained by fivefold cross‐validation. The Spearman r _s values are shown on the left, followed by the DSC_high, DSC_moderate, and DSC_low values between SPECT V and the two DL models. * p < 0.001. CNN, convolutional neural network; DSC, dice similarity coefficient; MCD, Monte Carlo dropout; n.s., not significant; SPECT V, single‐photon emission computed tomography ventilation

FIGURE 5

Typical cases of prediction of CTVI_{MCD U‐Net} in coronal slices. The anterior slice is shown on the top row, followed by the near, center, and posterior slices. The SPECT V is shown on the left, followed by CTVI_{MCD U‐Net}, the scaled uncertainty for the CTVI_{MCD U‐Net}, and masked CT. The ventilation values for viewing were normalized to give the 99th percentile value of 1 and 1st percentile value of 0. CT, computed tomography; CTVI, computed tomography ventilation imaging; MCD, Monte Carlo dropout; SPECT V, single‐photon emission computed tomography ventilation