Transmembrane pH gradient imaging in rodent glioma models.

A unique feature of the tumor microenvironment is extracellular acidosis in relation to intracellular milieu. Metabolic reprogramming in tumors results in overproduction of H+ ions (and lactate), which are extruded from the cells to support tumor survival and progression. As a result, the transmembrane pH gradient (ΔpH), representing the difference between intracellular pH (pHi) and extracellular pH (pHe), is posited to be larger in tumors compared with normal tissue. Controlling the transmembrane pH difference has promise as a potential therapeutic target in cancer as it plays an important role in regulating drug delivery into cells. The current study shows successful development of an MRI/MRSI-based technique that provides ΔpH imaging at submillimeter resolution. We applied this technique to image ΔpH in rat brains with RG2 and U87 gliomas, as well as in mouse brains with GL261 gliomas. pHi was measured with Amine and Amide Concentration-Independent Detection (AACID), while pHe was measured with Biosensor Imaging of Redundant Deviation in Shifts (BIRDS). The results indicate that pHi was slightly higher in tumors (7.40-7.43 in rats, 7.39-7.47 in mice) compared with normal brain (7.30-7.38 in rats, 7.32-7.36 in mice), while pHe was significantly lower in tumors (6.62-6.76 in rats, 6.74-6.84 in mice) compared with normal tissue (7.17-7.22 in rats, 7.20-7.21 in mice). As a result, ΔpH was higher in tumors (0.64-0.81 in rats, 0.62-0.65 in mice) compared with normal brain (0.13-0.16 in rats, 0.13-0.16 in mice). This work establishes an MRI/MRSI-based platform for ΔpH imaging at submillimeter resolution in gliomas.

Quantitative Automated Segmentation of Lipiodol Deposits on Cone-Beam CT Imaging Acquired during Transarterial Chemoembolization for Liver Tumors: A Deep Learning Approach.

PURPOSE: To show that a deep learning (DL)-based, automated model for Lipiodol (Guerbet Pharmaceuticals, Paris, France) segmentation on cone-beam computed tomography (CT) after conventional transarterial chemoembolization performs closer to the "ground truth segmentation" than a conventional thresholding-based model. MATERIALS AND METHODS: This post hoc analysis included 36 patients with a diagnosis of hepatocellular carcinoma or other solid liver tumors who underwent conventional transarterial chemoembolization with an intraprocedural cone-beam CT. Semiautomatic segmentation of Lipiodol was obtained. Subsequently, a convolutional U-net model was used to output a binary mask that predicted Lipiodol deposition. A threshold value of signal intensity on cone-beam CT was used to obtain a Lipiodol mask for comparison. The dice similarity coefficient (DSC), mean squared error (MSE), center of mass (CM), and fractional volume ratios for both masks were obtained by comparing them to the ground truth (radiologist-segmented Lipiodol deposits) to obtain accuracy metrics for the 2 masks. These results were used to compare the model versus the threshold technique. RESULTS: For all metrics, the U-net outperformed the threshold technique: DSC (0.65 ± 0.17 vs 0.45 ± 0.22, P < .001) and MSE (125.53 ± 107.36 vs 185.98 ± 93.82, P = .005). The difference between the CM predicted and the actual CM was 15.31 mm ± 14.63 versus 31.34 mm ± 30.24 (P < .001), with lesser distance indicating higher accuracy. The fraction of volume present ([predicted Lipiodol volume]/[ground truth Lipiodol volume]) was 1.22 ± 0.84 versus 2.58 ± 3.52 (P = .048) for the current model's prediction and threshold technique, respectively. CONCLUSIONS: This study showed that a DL framework could detect Lipiodol in cone-beam CT imaging and was capable of outperforming the conventionally used thresholding technique over several metrics. Further optimization will allow for more accurate, quantitative predictions of Lipiodol depositions intraprocedurally.