Utilization of artificial intelligence in minimally invasive right adrenalectomy: recognition of anatomical landmarks with deep learning.

BACKGROUND: The primary surgical approach for removing adrenal masses is minimally invasive adrenalectomy. Recognition of anatomical landmarks during surgery is critical for minimizing complications. Artificial intelligence-based tools can be utilized to create real-time navigation systems during laparoscopic and robotic right adrenalectomy. In this study, we aimed to develop deep learning models that can identify critical anatomical structures during minimally invasive right adrenalectomy. METHODS: In this experimental feasibility study, intraoperative videos of 20 patients who underwent minimally invasive right adrenalectomy in a tertiary care center between 2011 and 2023 were analyzed and used to develop an artificial intelligence-based anatomical landmark recognition system. Semantic segmentation of the liver, the inferior vena cava (IVC), and the right adrenal gland were performed. Fifty random images per patient during the dissection phase were extracted from videos. The experiments on the annotated images were performed on two state-of-the-art segmentation models named SwinUNETR and MedNeXt, which are transformer and convolutional neural network (CNN)-based segmentation architectures, respectively. Two loss function combinations, Dice-Cross Entropy and Dice-Focal Loss were experimented with for both of the models. The dataset was split into training and validation subsets with an 80:20 distribution on a patient basis in a 5-fold cross-validation approach. To introduce a sample variability to the dataset, strong-augmentation techniques were performed using intensity modifications and perspective transformations to represent different surgery environment scenarios. The models were evaluated by Dice Similarity Coefficient (DSC) and Intersection over Union (IoU) which are widely used segmentation metrics. For pixelwise classification performance, accuracy, sensitivity and specificity metrics were calculated on the validation subset. RESULTS: Out of 20 videos, 1000 images were extracted, and the anatomical landmarks (liver, IVC, and right adrenal gland) were annotated. Randomly distributed 800 images and 200 images were selected for the training and validation subsets, respectively. Our benchmark results show that the utilization of Dice-Cross Entropy Loss with the transformer-based SwinUNETR model achieved 78.37%, whereas the CNN-based MedNeXt model reached a 77.09% mDSC score. Conversely, MedNeXt reaches a higher mIoU score of 63.71% than SwinUNETR by 62.10% on a three-region prediction task. CONCLUSION: Artificial intelligence-based systems can predict anatomical landmarks with high performance in minimally invasive right adrenalectomy. Such tools can later be used to create real-time navigation systems during surgery in the near future.

Plasmonic Functional Assay Platform Determines the Therapeutic Profile of Cancer Cells.

Functional assay platforms could identify the biophysical properties of cells and their therapeutic response to drug treatments. Despite their strong ability to assess cellular pathways, functional assays require large tissue samples, long-term cell culture, and bulk measurements. Even though such a drawback is still valid, these limitations did not hinder the interest in these platforms for their capacity to reveal drug susceptibility. Some of the limitations could be overcome with single-cell functional assays by identifying subpopulations using small sample volumes. Along this direction, in this article, we developed a high-throughput plasmonic functional assay platform to identify the growth profile of cells and their therapeutic profile under therapies using mass and growth rate statistics of individual cells. Our technology could determine populations' growth profiles using the growth rate data of multiple single cells of the same population. Evaluating spectral variations based on the plasmonic diffraction field intensity images in real time, we could simultaneously monitor the mass change for the cells within the field of view of a camera with the capacity of > ∼500 cells/h scanning rate. Our technology could determine the therapeutic profile of cells under cancer drugs within few hours, while the classical techniques require days to show reduction in viability due to antitumor effects. The platform could reveal the heterogeneity within the therapeutic profile of populations and determine subpopulations showing resistance to drug therapies. As a proof-of-principle demonstration, we studied the growth profile of MCF-7 cells and their therapeutic behavior to standard-of-care drugs that have antitumor effects as shown in the literature, including difluoromethylornithine (DFMO), 5-fluorouracil (5-FU), paclitaxel (PTX), and doxorubicin (Dox). We successfully demonstrated the resistant behavior of an MCF-7 variant that could survive in the presence of DFMO. More importantly, we could precisely identify synergic and antagonistic effects of drug combinations based on the order of use in cancer therapy. Rapidly assessing the therapeutic profile of cancer cells, our plasmonic functional assay platform could be used to reveal personalized drug therapies for cancer patients.

Handheld plasmonic biosensor for virus detection in field-settings.

After World Health Organization (WHO) announced COVID-19 outbreak a pandemic, we all again realized the importance of developing rapid diagnostic kits. In this article, we introduced a lightweight and field-portable biosensor employing a plasmonic chip based on nanohole arrays integrated to a lensfree-imaging framework for label-free detection of viruses in field-settings. The platform utilizes a CMOS (complementary metal-oxide-semiconductor) camera with high quantum efficiency in the spectral window of interest to monitor diffraction field patterns of nanohole arrays under the uniform illumination of an LED (light-emitting diode) source which is spectrally tuned to the plasmonic mode supported by the nanohole arrays. As an example for the applicability of our biosensor for virus detection, we could successfully demonstrate the label-free detection of H1N1 viruses, e.g., swine flu, with medically relevant concentrations. We also developed a low-cost and easy-to-use sample preparation kit to prepare the surface of the plasmonic chip for analyte binding, e.g., virus-antibody binding. In order to reveal a complete biosensor technology, we also developed a user friendly Python™ - based graphical user interface (GUI) that allows direct access to biosensor hardware, taking and processing diffraction field images, and provides virus information to the end-user. Employing highly sensitive nanohole arrays and lensfree-imaging framework, our platform could yield an LOD as low as 103 TCID50/mL. Providing accurate and rapid sensing information in a handheld platform, weighing only 70 g and 12 cm tall, without the need for bulky and expensive instrumentation, our biosensor could be a very strong candidate for diagnostic applications in resource-poor settings. As our detection scheme is based on the use of antibodies, it could quickly adapt to the detection of different viral diseases, e.g., COVID-19 or influenza, by simply coating the plasmonic chip surface with an antibody possessing affinity to the virus type of interest. Possessing this ability, our biosensor could be swiftly deployed to the field in need for rapid diagnosis, which may be an important asset to prevent the spread of diseases before turning into a pandemic by isolating patients from the population.