Search

Image segmentation for 3D printing and 3D visualization has become an essential component in many fields of medical research, teaching, and clinical practice. Medical image segmentation requires sophisticated computerized quantifications and visualization tools. Recently, with the development of artificial intelligence (AI) technology, tumors or organs can be quickly and accurately detected and automatically contoured from medical images. This paper introduces a platform-independent, multi-modality image registration, segmentation, and 3D visualization program, named artificial intelligence-based medical image segmentation for 3D printing and naked eye 3D visualization (AIMIS3D). YOLOV3 algorithm was used to recognize prostate organ from T2-weighted MRI images with proper training. Prostate cancer and bladder cancer were segmented based on U-net from MRI images. CT images of osteosarcoma were loaded into the platform for the segmentation of lumbar spine, osteosarcoma, vessels, and local nerves for 3D printing. Breast displacement during each radiation therapy was quantitatively evaluated by automatically identifying the position of the 3D printed plastic breast bra. Brain vessel from multi-modality MRI images was segmented by using model-based transfer learning for 3D printing and naked eye 3D visualization in AIMIS3D platform.

Background: In the United States, prostate cancer has a relatively large impact on men's health. Magnetic resonance imaging (MRI) is useful for the diagnosis and treatment of prostate cancer. Introduction: The purpose of this study was to develop a quantitative marker for use in prostate cancer magnetization transfer (MT) magnetic resonance imaging (MRI) studies that is independent of radiofrequency (RF) saturation amplitude. Methods: Eighteen patients with biopsy-proven prostate cancer were enrolled in this study. MTMRI images were acquired using four RF saturation amplitudes at 33 frequency offsets. ROIs were delineated for the peripheral zone (PZ), central gland (CG), and tumor. Z-spectral data were collected in each region and fit to a three-parameter equation. The three parameters are: the magnitude of the bulk water pool (Aw), the full width at half maximum of the water pool (Gw), and the magnitude of the bound pool (Ab), while, the slopes from the linear regressions of Gw and Ab on RF saturation amplitude (called kAb and kGw) were used as quantitative markers. Results: A pairwise statistically significant difference was found between the PZ and tumor regions for the two saturation amplitude-independent quantitative markers. No pairwise statistically significant differences were found between the CG and tumor regions for any quantitative markers. Conclusion: The significant differences between the values of the two RF saturation amplitudeindependent quantitative markers in the PZ and tumor regions reveal that these markers may be capable of distinguishing healthy PZ tissue from prostate cancer.

Grain number per rice panicle, which directly determines grain yield, is an important agronomic trait for rice breeding and yield-related research. However, manually counting grains of rice per panicle is time-consuming, laborious, and error-prone. In this research, a grain detection model was proposed to automatically recognize and count grains on primary branches of a rice panicle. The model used image analysis based on deep learning convolutional neural network (CNN), by integrating the feature pyramid network (FPN) into the faster R-CNN network. The performance of the grain detection model was compared to that of the original faster R-CNN model and the SSD model, and it was found that the grain detection model was more reliable and accurate. The accuracy of the grain detection model was not affected by the lighting condition in which images of rice primary branches were taken. The model worked well for all rice branches with various numbers of grains. Through applying the grain detection model to images of fresh and dry branches, it was found that the model performance was not affected by the grain moisture conditions. The overall accuracy of the grain detection model was 99.4%. Results demonstrated that the model was accurate, reliable, and suitable for detecting grains of rice panicles with various conditions.

Prostate cancer is the second most common cancer in men worldwide. In the past few decades, multi-parameter MRI images have been used more and more in the detection of prostate cancer, and have important clinical significance in the invasive detection of prostate cancer. However, as the amount of data increases, radiologists need to devote more energy to recognizing images and simultaneously observing multiple sequences of magnetic resonance images, which is largely affected by the doctor's subjective judgment. Our goal is to build a new multi-parameter MRI-based prostate cancer tissue segmentation system to help doctors identify prostate cancer tissue from MRI data. We propose a convolutional neural network structure based on the U-net model to segment prostate cancer tissue. A neural network-based fusion algorithm is used to fuse the two-parameter MRI data, the fused image is sent to the convolutional neural network structure for training, and then the test set image is tested. Experimental results show that the algorithm can effectively segment prostate cancer tissue and has the potential for clinical application.

Prostate cancer is the second highest incidence of cancer in men worldwide. Therefore, early detection of prostate cancer is of great clinical significance. In the past few decades, multi-parametric MRI has been increasingly used in the detection of prostate cancer. However, this requires the radiologists to have a higher degree of prostate MRI reading experiences, which may be greatly influenced by the subjective judgment. Our goal is to establish a prostate cancer tissue segmentation system based on multiparametric MRI to help doctors automatically define prostate cancer regions. We proposed a convolutional neural network architecture based on the U-net model for segmentation of prostate cancer tissue. The multiparametric MRI images of the pelvis are rescaled, merged together, and delivered into a convolutional neural network architecture for training. Test images of the pelvis were used to evaluate the capability of the trained CNN architecture. The results showed that the proposed algorithm can efficiently segment prostate cancer tissue with a high accuracy and can be applied to computer assisted prostate cancer diagnosis.

The number of productive tillers per plant is one of the important agronomic traits associated with the grain yield of rice crop. However, manual counting of productive tillers is time-consuming, laborious and error-prone. In this study, a method for automatically detecting and counting productive tillers of rice crop was proposed based on deep learning convolutional neural network (CNN). The CNN model was trained using large amounts of in-field images taken by mobile phones from various varieties of rice crops under various environmental conditions. A Web app, integrating the trained CNN model and a Django server, was designed for fast and high-throughput detection of productive tillers. The performance of the Web app was evaluated for field-based practical applications. Results showed that the selected CNN model had a high precision and a fast detection rate. Through applying the Web app to 200 in-field images with 5 to 30 tillers per image, the number of productive tillers detected agreed well with manual counting data, regardless of rice variety or type of mobile phone used for image taking. The coefficients of determination between the Web app detection and manual counting of tillers were over 0.97 in all cases. Overall, compared to the manual counting, the accuracy of the Web app was over 99%. Furthermore, the performance of the Web app was not affected by the environmental conditions, such as illumination condition (cloudy or sunny) and water reflection in paddy fields.

Leaf color is correlated with nitrogen content, and detection of nitrogen content in rice leaves is important for guiding farmers in applying fertilizer. However, the performance of existing detection methods highly depends on the field environmental condition. Also, these methods require special imaging and computing equipment. To fill these gaps, a smartphone app was developed based on a standard leaf color chart (LCC) to detect color levels of rice leaves. Using the app developed, regions of rice leaf and LCC in an image were successfully identified by the color threshold segmentation. The color features of each region were effectively extracted using the CIELAB histograms. The color difference values between leaf and LCC that were calculated by the CIEDE2000 formula could be used to differentiate the color levels of rice leaves. The app was tested in field conditions. The results were accurate 96% of the times. Compared with manual inspections, the accuracies of the smartphone app in determining the color levels of rice leaves were higher than 92%. The average runtime for processing a leaf image in a field condition was 248 ms when using a Xiaomi Mi5 smartphone. The app also worked well after being implemented in other smartphones. The smartphone app allowed for an accurate, time-efficient, and low-cost detection of rice leaf color levels, which will help farmers in making decisions related to nitrogen fertilizer management for rice production.

Magnetic resonance elastography (MRE) can be used to noninvasively resolve the displacement pattern of induced mechanical waves propagating in tissue. The goal of this study is to establish an ergonomically flexible passive-driver design for brain MRE, to evaluate the reproducibility of MRE tissue-stiffness measurements, and to investigate the relationship between tissue-stiffness measurements and driver frequencies. An ergonomically flexible passive pillow-like driver was designed to induce mechanical waves in the brain. Two-dimensional finite-element simulation was used to evaluate mechanical wave propagation patterns in brain tissues. MRE scans were performed on 10 healthy volunteers at mechanical frequencies of 60, 50, and 40 Hz. An axial mid-brain slice was acquired using an echo-planar imaging sequence to map the displacement pattern with the motion-encoding gradient along the through-plane (z) direction. All subjects were scanned and rescanned within 1 h. The Wilcoxon signed-rank test was used to test for differences between white matter and gray matter shear-stiffness values. One-way analysis of variance (ANOVA) was used to test for differences between shear-stiffness measurements made at different frequencies. Scan-rescan reproducibility was evaluated by calculating the within-subject coefficient of variation (CV) for each subject. The finite-element simulation showed that a pillow-like passive driver is capable of efficient shear-wave propagation through brain tissue. No subjects complained about discomfort during MRE acquisitions using the ergonomically designed driver. The white-matter elastic modulus (mean ± standard deviation) across all subjects was 3.85 ± 0.12 kPa, 3.78 ± 0.15 kPa, and 3.36 ± 0.11 kPa at frequencies of 60, 50, and 40 Hz, respectively. The gray-matter elastic modulus across all subjects was 3.33 ± 0.14 kPa, 2.82 ± 0.16 kPa, and 2.24 ± 0.14 kPa at frequencies of 60, 50, and 40 Hz, respectively. The Wilcoxon signed-rank test confirmed that the shear stiffness was significantly higher in white matter than gray matter at all three frequencies. The ranges of within-subject coefficients of variation for white matter, gray matter, and whole-brain shear-stiffness measurements for the three frequencies were 1.8–3.5% (60 Hz), 4.7–6.0% (50 Hz), and 3.7–4.1% (40 Hz). An ergonomic pneumatic pillow-like driver is feasible for highly reproducible in vivo evaluation of brain-tissue shear stiffness. Brain-tissue shear-stiffness values were frequency-dependent, thus emphasizing the importance of standardizing MRE acquisition protocols in multi-center studies.

Based on the analysis about the main controlling factors of the formation and enrichment of lithologic hydrocarbon reservoirs of sandy conglomerate in the Bohai Bay Basin, and combined with qualitative and quantitative analysis, a classified evaluation mode was established including four factors that are tectonic setting, sedimentary environment, reservoirs and hydrocarbon source rocks. In the evaluation of geologic settings of one rift basin, the authors analyzed and compared the geologic conditions, hydrocarbon bearing and exploratory results of five types of structural zones and three types of sedimentary systems. In the reservoir evaluation, an evaluation item of the reservoir structure was added including six styles and evaluation grades. Combining the classified evaluation of the four factors with the routine analysis of exploration target areas, an evaluation standard of lithologic traps in rift basins was constructed, which is useful to evaluate lithologic targets in rift basins of eastern China. Key words: Bohai Bay Basin, rift basin, lithologic reservoir, sedimentary facies, reservoir structure, target area evaluation

Human linguistic units are hierarchical, and our brain responds differently when processing linguistic units during sentence comprehension, especially when the modality of the received signal is different (auditory, visual, or audio-visual). However, it is unclear how the brain processes and integrates language information at different linguistic units (words, phrases, and sentences) provided simultaneously in audio and visual modalities. To address the issue, we presented participants with sequences of short Chinese sentences through auditory or visual or combined audio- visual modalities, while electroencephalographic responses were recorded. With a frequency tagging approach, we analyzed the neural representations of basic linguistic units (i.e., characters/monosyllabic words) and higher-level linguistic structures (i.e., phrases and sentences) across the three modalities separately. We found that audio-visual integration occurs at all linguistic units, and the brain areas involved in the integration varied across different linguistic levels. In particular, the integration of sentences activated the local left prefrontal area. Therefore, we used continuous theta-burst stimulation (cTBS) to verify that the left prefrontal cortex plays a vital role in the audio-visual integration of sentence information. Our findings suggest the advantage of bimodal language comprehension at hierarchical stages in language-related information processing and provide evidence for the causal role of the left prefrontal regions in processing information of audio-visual sentences.