Predicting treatment response in multicenter non-small cell lung cancer patients based on federated learning.

BACKGROUND: Multicenter non-small cell lung cancer (NSCLC) patient data is information-rich. However, its direct integration becomes exceptionally challenging due to constraints involving different healthcare organizations and regulations. Traditional centralized machine learning methods require centralizing these sensitive medical data for training, posing risks of patient privacy leakage and data security issues. In this context, federated learning (FL) has attracted much attention as a distributed machine learning framework. It effectively addresses this contradiction by preserving data locally, conducting local model training, and aggregating model parameters. This approach enables the utilization of multicenter data with maximum benefit while ensuring privacy safeguards. Based on pre-radiotherapy planning target volume images of NSCLC patients, a multicenter treatment response prediction model is designed by FL for predicting the probability of remission of NSCLC patients. This approach ensures medical data privacy, high prediction accuracy and computing efficiency, offering valuable insights for clinical decision-making. METHODS: We retrospectively collected CT images from 245 NSCLC patients undergoing chemotherapy and radiotherapy (CRT) in four Chinese hospitals. In a simulation environment, we compared the performance of the centralized deep learning (DL) model with that of the FL model using data from two sites. Additionally, due to the unavailability of data from one hospital, we established a real-world FL model using data from three sites. Assessments were conducted using measures such as accuracy, receiver operating characteristic curve, and confusion matrices. RESULTS: The model's prediction performance obtained using FL methods outperforms that of traditional centralized learning methods. In the comparative experiment, the DL model achieves an AUC of 0.718/0.695, while the FL model demonstrates an AUC of 0.725/0.689, with real-world FL model achieving an AUC of 0.698/0.672. CONCLUSIONS: We demonstrate that the performance of a FL predictive model, developed by combining convolutional neural networks (CNNs) with data from multiple medical centers, is comparable to that of a traditional DL model obtained through centralized training. It can efficiently predict CRT treatment response in NSCLC patients while preserving privacy.

Downregulation of Rad51 Expression and Activity Potentiates the Cytotoxic Effect of Osimertinib in Human Non-Small Cell Lung Cancer Cells.

INTRODUCTION: Osimertinib (AZD9291) is a third-generation epidermal growth factor receptor (EGFR) tyrosine kinase inhibitor that has shown significant clinical benefits in patients with EGFR-sensitizing mutations or the EGFR T790M mutation. The homologous recombination (HR) pathway is crucial for repairing DNA double-strand breaks (DSBs). Rad51 plays a central role in HR, facilitating the search for homology and promoting DNA strand exchange between homologous DNA molecules. Rad51 is overexpressed in numerous types of cancer cells. B02, a specific small molecule inhibitor of Rad51, inhibits the DNA strand exchange activity of Rad51. Previous studies have indicated that B02 disrupted Rad51 foci formation in response to DNA damage and inhibited DSBs repair in human cells and sensitized them to chemotherapeutic drugs in vitro and in vivo. However, the potential therapeutic effects of combining osimertinib with a Rad51 inhibitor are not well understood. The aim of this study was to elucidate whether the downregulation of Rad51 expression and activity can enhance the osimertinib-induced cytotoxicity in non-small cell lung cancer (NSCLC) cells. METHODS: We used the MTS, trypan blue dye exclusion and colony-formation ability assay to determine whether osimertinib alone or in combination with B02 had cytotoxic effects on NSCLC cell lines. Real-time polymerase chain reaction was conducted to measure the amounts of Rad51 mRNA. The protein levels of phosphorylated AKT and Rad51 were determined by Western blot analysis. RESULTS: We found that osimertinib reduced Rad51 expression by inactivating AKT activity. Rad51 knockdown using small interfering RNA or AKT inactivation through the phosphatidylinositol 3-kinase inhibitor LY294002 or si-AKT RNA transfection enhanced the cytotoxic and growth inhibitory effects of osimertinib. In contrast, AKT-CA (a constitutively active form of AKT) vector-enforced expression could mitigate the cytotoxic and cell growth inhibitory effects of osimertinib. Furthermore, B02 significantly enhanced the cytotoxic and cell growth inhibitory effects of osimertinib in NSCLC cells. Compared to parental cells, the activation of AKT and Rad51 expression in osimertinib-resistant cells could not be significantly inhibited by osimertinib treatment. Moreover, the increased expression of Rad51 is associated with the resistance mechanism in osimertinib-resistant H1975 and A549 cells. CONCLUSION: Collectively, the downregulation of Rad51 expression and activity enhances the cytotoxic effect of osimertinib in human NSCLC cells.

Iterative Reconstruction of Micro Computed Tomography Scans Using Multiple Heterogeneous GPUs.

Graphics processing units (GPUs) facilitate massive parallelism and high-capacity storage, and thus are suitable for the iterative reconstruction of ultrahigh-resolution micro computed tomography (CT) scans by on-the-fly system matrix (OTFSM) calculation using ordered subsets expectation maximization (OSEM). We propose a finite state automaton (FSA) method that facilitates iterative reconstruction using a heterogeneous multi-GPU platform through parallelizing the matrix calculations derived from a ray tracing system of ordered subsets. The FSAs perform flow control for parallel threading of the heterogeneous GPUs, which minimizes the latency of launching ordered-subsets tasks, reduces the data transfer between the main system memory and local GPU memory, and solves the memory-bound of a single GPU. In the experiments, we compared the operation efficiency of OS-MLTR for three reconstruction environments. The heterogeneous multiple GPUs with job queues for high throughput calculation speed is up to five times faster than the single GPU environment, and that speed up is nine times faster than the heterogeneous multiple GPUs with the FIFO queues of the device scheduling control. Eventually, we proposed an event-triggered FSA method for iterative reconstruction using multiple heterogeneous GPUs that solves the memory-bound issue of a single GPU at ultrahigh resolutions, and the routines of the proposed method were successfully executed on each GPU simultaneously.

Comparison of Calcium Scoring Between 64-Multidetector Computed Tomography and 320-Multidetector Computed Tomography Using a Cardiac Phantom: Achieving Consistent Image Quality With Dose Optimization.

ABSTRACT: The purpose of this study was to evaluate the relationship between radiation dose and noise level on various coronary calcium scoring protocols between 64-multidetector computed tomography (MDCT) and 320-MDCT. The cardiac QRM phantoms (1 small size and 1 medium size) were used in this study. Lower-dose imaging protocols were proposed for optimization with the parameters of 120 kVp and 10 mAs for small-size phantom (0.336 mSv) in 64-MDCT imaging and small-size phantom (0.2 mSv) in 320-MDCT case, and 120 kVp and 80 mAs for medium-size phantom (2.73 mSv) in 64-MDCT imaging and medium-size phantom (1.58 mSv) in 320-MDCT case. Our results suggest that people can apply lower-dose protocols in the clinical use for early diagnosis of coronary disease without sacrificing diagnostic accuracy.

Downregulation of Xeroderma Pigmentosum Complementation Group C Expression by 17-Allylamino-17-Demethoxygeldanamycin Enhances Bevacizumab-Induced Cytotoxicity in Human Lung Cancer Cells.

INTRODUCTION: Xeroderma pigmentosum complementation group C (XPC) protein is an important DNA damage recognition factor involved in nucleotide excision repair and regulation of non-small-cell lung cancer (NSCLC) cell proliferation and viability. 17-Allylamino-17-demethoxygeldanamycin (17-AAG) blocks ATP binding to heat shock protein 90 (Hsp90), resulting in destabilization of Hsp90-client protein complexes. Vascular endothelial growth factor (VEGF) is a potent angiogenic growth factor expressed by many types of tumors. Bevacizumab (Avastin) is a humanized monoclonal antibody against human VEGF used as an antiangiogenesis agent in the therapy of many cancers, proving successful in increasing objective tumor response rate and prolonging overall survival in NSCLC patients. METHODS: After the bevacizumab and/or 17-AAG treatment, the expressions of XPC mRNA were determined by quantitative real-time PCR analysis. Protein levels of XPC and phospho-AKT were determined by Western blot analysis. We used specific XPC small interfering RNA and PI3K inhibitor (LY294002) to examine the role of the AKT-XPC signal in regulating the chemosensitivity of bevacizumab and 17-AAG. Cell viability was assessed by the MTS assay and trypan blue exclusion assay. RESULTS: In this study, bevacizumab decreased XPC expression in human lung squamous cell carcinoma H520 and H1703 cells via AKT inactivation. Enhancement of AKT activity by transfection with constitutively active AKT vectors increased XPC expression and cell survival after treatment with bevacizumab. In addition, 17-AAG synergistically enhanced bevacizumab-induced cytotoxicity and cell growth inhibition in H520 and H1703 cells, associated with downregulation of XPC expression and inactivation of AKT. DISCUSSION/CONCLUSION: Together, these results may provide a rationale to combine bevacizumab with Hsp90 inhibitors in future to enhance therapeutic effects for lung cancer.

Downregulation of p38 MAPK Activation and Radiation-Sensitive 52 Expression Enhances 5-Fluorouracil and Erlotinib-Induced Cytotoxicity in Human Lung Squamous Cells.

INTRODUCTION: 5-Fluorouracil (5-FU) is used to treat various cancers, including non-small-cell lung cancer (NSCLC). It inhibits nucleotide synthesis and induces single- and double-strand DNA breaks. In the homologous recombination pathway, radiation-sensitive 52 (Rad52) plays a crucial role in DNA repair by promoting the annealing of complementary single-stranded DNA and stimulating Rad51 recombinase activity. Erlotinib (Tarceva) is a selective epidermal growth factor receptor tyrosine kinase inhibitor with clinical activity against NSCLC cells. However, whether the combination of 5-FU and erlotinib has synergistic activity against NSCLC cells is unknown. METHODS: After the 5-FU and/or erlotinib treatment, the expressions of Rad52 mRNA were determined by quantitative real-time polymerase chain reaction analysis. Protein levels of Rad52 and phospho-p38 MAPK were determined by Western blot analysis. We used specific Rad52 or p38 MAPK small interfering RNA and p38 MAPK inhibitor (SB2023580) to examine the role of p38 MAPK-Rad52 signal in regulating the chemosensitivity of 5-FU and/or erlotinib. Cell viability was assessed by MTS assay and trypan blue exclusion assay. RESULTS: In 2 squamous cell carcinoma cell lines, namely, H520 and H1703, 5-FU reduced Rad52 expression in a p38 MAPK inactivation-dependent manner. Enhancement of p38 MAPK activity by transfection with MKK6E (a constitutively active form of MKK6) vector increased the Rad52 protein level and cell survival by 5-FU. However, in human lung bronchioloalveolar cell adenocarcinoma A549 cells, 5-FU reduced Rad52 expression and induced cytotoxicity independent of p38 MAPK. Moreover, 5-FU synergistically enhanced the cytotoxicity and cell growth inhibition of erlotinib in NSCLC cells; these effects were associated with Rad52 downregulation and p38 MAPK inactivation in H520 and H1703 cells. CONCLUSION: The results provide a rationale for combining 5-FU and erlotinib in lung cancer treatment.

Nitroglycerin Enhances Cisplatin-Induced Cytotoxicity via AKT Inactivation and Thymidylate Synthase Downregulation in Human Lung Cancer Cells.

Nitroglycerin (NTG), a nitric oxide-donating drug, may increase tumor blood flow and consequently increase cancer drug delivery to tumor cells. Thymidylate synthase (TS) is an essential enzyme for the de novo synthesis of deoxythymidine monophosphate; we had found that knocking down the expression of TS sensitizes lung cancer cells to cisplatin-induced cytotoxicity. However, whether NTG and cisplatin could induce synergistic cytotoxicity in nonsmall cell lung cancer (NSCLC) cells through modulating TS expression is unknown. In this study, NTG decreased TS expression in an AKT, also known as Protein kinase B (PKB) inactivation dependent manner in human lung adenocarcinoma A549 and squamous cell carcinoma H1703 cells. Enhancement of AKT activity by transfection with constitutive active AKT vectors increased the TS expression level as well as the cell survival pretreated by NTG. Moreover, NTG synergistically enhanced cytotoxicity and cell growth inhibition by cisplatin treatment in NSCLC cells, which were associated with downregulation of TS expression and inactivation of AKT in A549 and H1703 cells. Together, these results may provide a rationale to combine NTG with cisplatin-based chemotherapy to enhance the therapeutic effect for lung cancer in the future.

Salinomycin acts through reducing AKT-dependent thymidylate synthase expression to enhance erlotinib-induced cytotoxicity in human lung cancer cells.

Erlotinib (TarcevaR) is a selective epidermal growth factor receptor (EGFR) tyrosine kinase inhibitor in the treatment of human non-small cell lung cancer (NSCLC). Salinomycin, a polyether antibiotic, has been promising a novel therapeutic agent for lung cancer, and down-regulated the expression of thymidylate synthase (TS) in NSCLC cell lines. Previous study showed that against EGFR and TS was strongly synergistic cytotoxicity in NSCLC cells. In this study, we showed that erlotinib (1.25-10µM) treatment down-regulating of TS expression in an AKT inactivation manner in two NSCLC cell lines, human lung squamous cell carcinoma H1703 and adenocarcinoma H1975 cells. Knockdown of TS using small interfering RNA (siRNA) or inhibiting AKT activity with PI3K inhibitor LY294002 enhanced the cytotoxicity and cell growth inhibition of erlotinib. A combination of erlotinib and salinomycin resulted in synergistic enhancement of cytotoxicity and cell growth inhibition in NSCLC cells, accompanied with reduced protein levels of phospho-AKT(Ser473), phospho-AKT(Thr308), and TS. Overexpression of a constitutive active AKT (AKT-CA) or Flag-TS expression vector reversed the salinomycin and erlotinib-induced synergistic cytotoxicity. Our findings suggested that the down-regulation of AKT-mediated TS expression by salinomycin enhanced the erlotinib-induced cytotoxicity in NSCLC cells. These results may provide a rationale to combine salinomycin with erlotinib for lung cancer treatment.

17-(Allylamino)-17-Demethoxygeldanamycin Enhances Etoposide-Induced Cytotoxicity via the Downregulation of Xeroderma Pigmentosum Complementation Group C Expression in Human Lung Squamous Cell Carcinoma Cells.

Etoposide (VP16) is a topoisomerase II inhibitor and has been used for the treatment of non-small cell lung cancer (NSCLC). Xeroderma pigmentosum complementation group C (XPC) protein is a DNA damage recognition factor in nucleotide excision repair and involved in regulating NSCLC cell proliferation and viability. Heat shock protein 90 (Hsp90) is a ubiquitous molecular chaperone that is responsible for the stabilization and maturation of many oncogenic proteins. In this study, we report whether Hsp90 inhibitor 17-allylamino-17-demethoxygeldanamycin (17-AAG) enhanced etoposide-induced cytotoxicity in NSCLC cells through modulating the XPC expression. We found that etoposide increased XPC expression in an AKT activation manner in 2 squamous cell carcinoma H1703 and H520 cells. Knockdown of XPC using siRNA or inactivation of AKT by pharmacological inhibitor PI3K inhibitor (LY294002) enhanced the cytotoxic effects of etoposide. In contrast, enforced expression of XPC cDNA or AKT-CA (a constitutively active form of AKT) reduced the cytotoxicity and cell growth inhibition of etoposide. Hsp90 inhibitor 17-AAG enhanced cytotoxicity and cell growth inhibition of etoposide in NSCLC cells, which were associated with the downregulation of XPC expression and inactivation of AKT. Our findings suggested that the Hsp90 inhibition induced XPC downregulation involved in enhancing the etoposide-induced cytotoxicity in H1703 and H520 cells.

Development of Limited-Angle Iterative Reconstruction Algorithms with Context Encoder-Based Sinogram Completion for Micro-CT Applications.

Limited-angle iterative reconstruction (LAIR) reduces the radiation dose required for computed tomography (CT) imaging by decreasing the range of the projection angle. We developed an image-quality-based stopping-criteria method with a flexible and innovative instrument design that, when combined with LAIR, provides the image quality of a conventional CT system. This study describes the construction of different scan acquisition protocols for micro-CT system applications. Fully-sampled Feldkamp (FDK)-reconstructed images were used as references for comparison to assess the image quality produced by these tested protocols. The insufficient portions of a sinogram were inpainted by applying a context encoder (CE), a type of generative adversarial network, to the LAIR process. The context image was passed through an encoder to identify features that were connected to the decoder using a channel-wise fully-connected layer. Our results evidence the excellent performance of this novel approach. Even when we reduce the radiation dose by 1/4, the iterative-based LAIR improved the full-width half-maximum, contrast-to-noise and signal-to-noise ratios by 20% to 40% compared to a fully-sampled FDK-based reconstruction. Our data support that this CE-based sinogram completion method enhances the efficacy and efficiency of LAIR and that would allow feasibility of limited angle reconstruction.

Astaxanthin enhances pemetrexed-induced cytotoxicity by downregulation of thymidylate synthase expression in human lung cancer cells.

Pemetrexed, a multitargeted antifolate agent, has demonstrated clinical activity in non-small cell lung cancer (NSCLC) cells. Increased expression of thymidylate synthase (TS) is thought to be associated with resistance to pemetrexed. Astaxanthin exhibits a wide range of beneficial effects including anti-cancer and anti-inflammatory properties. In this study, we showed that down-regulating of TS expression in two NSCLC cell lines, human lung adenocarcinoma H1650 and squamous cell carcinoma H1703 cells, with astaxanthin were associated with decreased MKK1/2-ERK1/2 activity. Enforced expression of constitutively active MKK1 (MKK1-CA) vector significantly rescued the decreased TS mRNA and protein levels in astaxanthin-treated NSCLC cells. Combined treatment with a MKK1/2 inhibitor (U0126 or PD98059) further decreased the TS expression in astaxanthin-exposed NSCLC cells. Knockdown of TS using small interfering RNA (siRNA) or inhibiting ERK1/2 activity enhanced the cytotoxicity and cell growth inhibition of astaxanthin. Combination of pemetrexed and astaxanthin resulted in synergistic enhancing cytotoxicity and cell growth inhibition in NSCLC cells, accompanied with reduced activation of phospho-MKK1/2, phopho-ERK1/2, and TS expression. Overexpression of MKK1/2-CA reversed the astaxanthin and pemetrexed-induced synergistic cytotoxicity. Our findings suggested that the down-regulation of MKK1/2-ERK1/2-mediated TS expression by astaxanthin is an important regulator of enhancing the pemetrexed-induced cytotoxicity in NSCLC cells.

Quantitative assessment of cerebral glucose metabolic rates after blood-brain barrier disruption induced by focused ultrasound using FDG-MicroPET.

The goal of this study was to evaluate the pharmacokinetics of (18)F-2-fluoro-2-deoxy-d-glucose ((18)F-FDG) and the expression of glucose transporter 1 (GLUT1) protein after blood-brain barrier (BBB) disruption of normal rat brains by focused ultrasound (FUS). After delivery of an intravenous bolus of ~37 MBq (1 mCi) (18)F-FDG, dynamic positron emission tomography scans were performed on rats with normal brains and those whose BBBs had been disrupted by FUS. Arterial blood sampling was collected throughout the scanning procedure. A 2-tissue compartmental model was used to estimate (18)F-FDG kinetic parameters in brain tissues. The rate constants Ki, K1, and k3 were assumed to characterize the uptake, transport, and hexokinase activity, respectively, of (18)F-FDG. The uptake of (18)F-FDG in brains significantly decreased immediately after the blood-brain barrier was disrupted. At the same time, the derived values of Ki, K1, and k3 for the sonicated brains were significantly lower than those for the control brains. In agreement with the reduction in glucose, Western blot analyses confirmed that focused ultrasound exposure significantly reduced the expression of GLUT1 protein in the brains. Furthermore, the effect of focused ultrasound on glucose uptake was transient and reversible 24h after sonication. Our results indicate that focused ultrasound may inhibit GLUT1 expression to decrease the glucose uptake in brain tissue during the period of BBB disruption.

Utilizing deep learning techniques to improve image quality and noise reduction in preclinical low-dose PET images in the sinogram domain.

BACKGROUND: Low-dose positron emission tomography (LD-PET) imaging is commonly employed in preclinical research to minimize radiation exposure to animal subjects. However, LD-PET images often exhibit poor quality and high noise levels due to the low signal-to-noise ratio. Deep learning (DL) techniques such as generative adversarial networks (GANs) and convolutional neural network (CNN) have the capability to enhance the quality of images derived from noisy or low-quality PET data, which encodes critical information about radioactivity distribution in the body. PURPOSE: Our objective was to optimize the image quality and reduce noise in preclinical PET images by utilizing the sinogram domain as input for DL models, resulting in improved image quality as compared to LD-PET images. METHODS: A GAN and CNN model were utilized to predict high-dose (HD) preclinical PET sinograms from the corresponding LD preclinical PET sinograms. In order to generate the datasets, experiments were conducted on micro-phantoms, animal subjects (rats), and virtual simulations. The quality of DL-generated images was weighted by performing the following quantitative measures: structural similarity index measure (SSIM), root mean squared error (RMSE), peak signal-to-noise ratio (PSNR), signal-to-noise ratio (SNR), and contrast-to-noise ratio (CNR). Additionally, DL input and output were both subjected to a spatial resolution calculation of full width half maximum (FWHM) and full width tenth maximum (FWTM). DL outcomes were then compared with the conventional denoising algorithms such as non-local means (NLM), block-matching, and 3D filtering (BM3D). RESULTS: The DL models effectively learned image features and produced high-quality images, as reflected in the quantitative metrics. Notably, the FWHM and FWTM values of DL PET images exhibited significantly improved accuracy compared to LD, NLM, and BM3D PET images, and just as precise as HD PET images. The MSE loss underscored the excellent performance of the models, indicating that the models performed well. To further improve the training, the generator loss (G loss) was increased to a value higher than the discriminator loss (D loss), thereby achieving convergence in the GAN model. CONCLUSIONS: The sinograms generated by the GAN network closely resembled real HD preclinical PET sinograms and were more realistic than LD. There was a noticeable improvement in image quality and noise factor in the predicted HD images. Importantly, DL networks did not fully compromise the spatial resolution of the images.

A simple and precise alignment calibration method for cone-beam computed tomography with the verifications.

Cone-beam Computed Tomography (CBCT) is widely used in dental imaging, small animal imaging, radiotherapy, and non-destructive industrial inspection. The quality of CBCT images depends on the precise knowledge of the CBCT system's alignment. We introduce a distinct procedure, "precision alignment loop (PAL)", to calibrate any CBCT system with a circular trajectory. We describe the calibration procedure by using a line-beads phantom, and how PAL determines the misalignments from a CBCT system. PAL also yields the uncertainties in the simulated calibration to give an estimate of the errors in the misalignments. From the analytical simulations, PAL can precisely obtain the source-to-rotation axis distance (SRD), and the geometric center G, "the point in z-axis meets the detector", where the z-axis is coincident with the line from the X-ray source that intersects the axis of the rotation (AOR) orthogonally. The uncertainties of three misalignment angles of the detector are within ±0.05°, which is close to ±0.04° for the results of Yang et al. [18], but our method is easy and simple to implement. Our distinct procedure, on the other hand, yields the calibration of a micro-CT system and an example of reconstructed images, showing our calibration method for the CBCT system to be simple, precise, and accurate.

Cone-beam x-ray luminescence computed tomography (CB-XLCT) prototype development and performance evaluation.

This study developed a prototype for a rotational cone-beam x-ray luminescence computed tomography (CB-XLCT) system, considering its potential application in pre-clinical theranostic imaging. A geometric calibration method applicable to both imaging chains (XL and CT) was also developed to enhance image quality. The results of systematic performance evaluations were presented to assess the feasibility of commercializing XLCT technology. Monte Carlo GATE simulation was performed to determine the optimal imaging conditions for nanophosphor particles (NPs) irradiated by 70 kV x-rays. We acquired a low-dose transmission x-ray tube and designed a prone positioning platform and a rotating gantry, using mice as targets from commercial small animalµ-CT systems. We then employed the image cross-correlation (ICC) automatic geometric calibration method to calibrate XL and CT images. The performance of the system was evaluated through a series of phantom experiments with a linearity of 0.99, and the contrast-to-noise ratio (CNR) between hydroxyl-apatite (HA) and based epoxy resin is 19.5. The XL images of the CB-XLCT prototype achieved a Dice similarity coefficient (DICE) of 0.149 for a distance of 1 mm between the two light sources. Finally, the final XLCT imaging results were demonstrated using the Letter phantoms with NPs. In summary, the CB-XLCT prototype developed in this study showed the potential to achieve high-quality imaging with acceptable radiation doses for small animals. The performance of CT images was comparable to current commercial machines, while the XL images exhibited promising results in phantom imaging, but further efforts are needed for biomedical applications.

Whole slide imaging-based deep learning to predict the treatment response of patients with non-small cell lung cancer.

Background: This study developed and validated a deep learning (DL) model based on whole slide imaging (WSI) for predicting the treatment response to chemotherapy and radiotherapy (CRT) among patients with non-small cell lung cancer (NSCLC). Methods: We collected the WSI of 120 nonsurgical patients with NSCLC treated with CRT from three hospitals in China. Based on the processed WSI, two DL models were established: a tissue classification model which was used to select tumor-tiles, and another model which predicted the treatment response of the patients based on the tumor-tiles (predicting the treatment response of each tile). A voting method was employed, by which the label of tiles with the greatest quantity from 1 patient would be used as the label of the patient. Results: The tissue classification model had a great performance (accuracy in the training set/internal validation set =0.966/0.956). Based on 181,875 tumor-tiles selected by the tissue classification model, the model for predicting the treatment response demonstrated strong predictive ability (accuracy of patient-level prediction in the internal validation set/external validation set 1/external validation set 2 =0.786/0.742/0.737). Conclusions: A DL model was constructed based on WSI to predict the treatment response of patients with NSCLC. This model can help doctors to formulate personalized CRT plans and improve treatment outcomes.

Effects of food hardness on temporomandibular joint osteoarthritis: Qualitative and quantitative micro-CT analysis of rats in vivo.

BACKGROUND: Temporomandibular joint osteoarthritis (TMJ-OA) is a degenerative joint disease in which quantitative analysis based on magnetic resonance image (MRI) or cone-beam computed tomography (CBCT) remains limited. Moreover, the long-term effects of soft food on the adaptive condylar remodeling process in TMJ-OA remain unclear. This study aimed to assess the effects of food hardness on adaptive condylar remodeling in a healthy TMJ, TMJ-OA, and controlled TMJ-OA. METHODS: Complete Freund's adjuvant (CFA) was used for TMJ-OA induction and Link-N (LN) for TMJ repair. Eighteen mature rats were randomly divided into six groups: (1) control/normal diet (Ctrl-N); (2) control/soft diet (Ctrl-S); (3) TMJ-OA/normal diet (CFA-N); (4) TMJ-OA/soft diet (CFA-S); (5) Link-N-controlled TMJ-OA/normal diet (LN-N); and (6) Link-N-controlled TMJ-OA/soft diet (LN-S). Micro-CT was performed 14, 21, and 28 days after CFA injection to analyze the bone volume, bone volume fraction (BVF), bone mineral density (BMD), and trabecular bone number and thickness (Tb.N, Tb.Th). MRI and histological imaging were performed to support the analysis. RESULTS: Under CFA treatment, the BVF and BMD decreased significantly (p < 0.01) and later recovered to normal. However, more significant improvements occurred in normal-diet groups than soft-diet groups. Additionally, bone volume changes were more predictable in the normal-diet groups than in the soft-diet groups. The normal-diet groups presented a significant decrease and increase in the Tb.N and Tb.Th, respectively (p < 0.05), while the Tb.N and Tb.Th in the soft-diet groups remained largely unchanged. Furthermore, a significantly higher frequency of irregularities on the condylar articular surface was found in the soft-diet groups. CONCLUSIONS: Compared with a soft diet, a normal diet may be beneficial for preserving condyle articular surface and directing bone remodeling in TMJ-OA rats.

Extraction of an input function from dynamic micro-PET images using wavelet packet based sub-band decomposition independent component analysis.

Positron emission tomography (PET) can be used to quantify physiological parameters. However to perform quantification requires that an input function is measured, namely a plasma time activity curve (TAC). Image-derived input functions (IDIFs) are attractive because they are noninvasive and nearly no blood loss is involved. However, the spatial resolution and the signal to noise ratio (SNR) of PET images are low, which degrades the accuracy of IDIFs. The objective of this study was to extract accurate input functions from microPET images with zero or one plasma sample using wavelet packet based sub-band decomposition independent component analysis (WP SDICA). Two approaches were used in this study. The first was the use of simulated dynamic rat images with different spatial resolutions and SNRs, and the second was the use of dynamic images of eight Sprague-Dawley rats. We also used a population-based input function and a fuzzy c-means clustering approach and compared their results with those obtained by our method using normalized root mean square errors, area under curve errors, and correlation coefficients. Our results showed that the accuracy of the one-sample WP SDICA approach was better than the other approaches using both simulated and realistic comparisons. The errors in the metabolic rate, as estimated by one-sample WP SDICA, were also the smallest using our approach.

Segmentation of 3D microPET images of the rat brain via the hybrid gaussian mixture method with kernel density estimation.

Segmentation of positron emission tomography (PET) is typically achieved using the K-Means method or other approaches. In preclinical and clinical applications, the K-Means method needs a prior estimation of parameters such as the number of clusters and appropriate initialized values. This work segments microPET images using a hybrid method combining the Gaussian mixture model (GMM) with kernel density estimation. Segmentation is crucial to registration of disordered 2-deoxy-2-fluoro-D-glucose (FDG) accumulation locations with functional diagnosis and to estimate standardized uptake values (SUVs) of region of interests (ROIs) in PET images. Therefore, simulation studies are conducted to apply spherical targets to evaluate segmentation accuracy based on Tanimoto's definition of similarity. The proposed method generates a higher degree of similarity than the K-Means method. The PET images of a rat brain are used to compare the segmented shape and area of the cerebral cortex by the K-Means method and the proposed method by volume rendering. The proposed method provides clearer and more detailed activity structures of an FDG accumulation location in the cerebral cortex than those by the K-Means method.

Virtual and real-world implementation of deep-learning-based image denoising model on projection domain in digital tomosynthesis and cone-beam computed tomography data.

Reducing the radiation dose will cause severe image noise and artifacts, and degradation of image quality will also affect the accuracy of diagnosis. To find a solution, we comprise a 2D and 3D concatenating convolutional encoder-decoder (CCE-3D) and the structural sensitive loss (SSL), via transfer learning (TL) denoising in the projection domain for low-dose computed tomography (LDCT), radiography, and tomosynthesis. The simulation and real-world practicing results show that many of the figures-of-merit (FOMs) increase in both projections (2-3 times) and CT imaging (1.5-2 times). From the PSNR and structural similarity index of measurement (SSIM), the CCE-3D model is effective in denoising but keeps the shape of the structure. Hence, we have developed a denoising model that can be served as a promising tool to be implemented in the next generation of x-ray radiography, tomosynthesis, and LDCT systems.