Parotid Gland Segmentation Using Purely Transformer-Based U-Shaped Network and Multimodal MRI.

Parotid gland tumors account for approximately 2% to 10% of head and neck tumors. Segmentation of parotid glands and tumors on magnetic resonance images is essential in accurately diagnosing and selecting appropriate surgical plans. However, segmentation of parotid glands is particularly challenging due to their variable shape and low contrast with surrounding structures. Recently, deep learning has developed rapidly, and Transformer-based networks have performed well on many computer vision tasks. However, Transformer-based networks have yet to be well used in parotid gland segmentation tasks. We collected a multi-center multimodal parotid gland MRI dataset and implemented parotid gland segmentation using a purely Transformer-based U-shaped segmentation network. We used both absolute and relative positional encoding to improve parotid gland segmentation and achieved multimodal information fusion without increasing the network computation. In addition, our novel training approach reduces the clinician's labeling workload by nearly half. Our method achieved good segmentation of both parotid glands and tumors. On the test set, our model achieved a Dice-Similarity Coefficient of 86.99%, Pixel Accuracy of 99.19%, Mean Intersection over Union of 81.79%, and Hausdorff Distance of 3.87. The purely Transformer-based U-shaped segmentation network we used outperforms other convolutional neural networks. In addition, our method can effectively fuse the information from multi-center multimodal MRI dataset, thus improving the parotid gland segmentation.

Swin MoCo: Improving parotid gland MRI segmentation using contrastive learning.

BACKGROUND: Segmentation of the parotid glands and tumors by MR images is essential for treating parotid gland tumors. However, segmentation of the parotid glands is particularly challenging due to their variable shape and low contrast with surrounding structures. PURPOSE: The lack of large and well-annotated datasets limits the development of deep learning in medical images. As an unsupervised learning method, contrastive learning has seen rapid development in recent years. It can better use unlabeled images and is hopeful to improve parotid gland segmentation. METHODS: We propose Swin MoCo, a momentum contrastive learning network with Swin Transformer as its backbone. The ImageNet supervised model is used as the initial weights of Swin MoCo, thus improving the training effects on small medical image datasets. RESULTS: Swin MoCo trained with transfer learning improves parotid gland segmentation to 89.78% DSC, 85.18% mIoU, 3.60 HD, and 90.08% mAcc. On the Synapse multi-organ computed tomography (CT) dataset, using Swin MoCo as the pre-trained model of Swin-Unet yields 79.66% DSC and 12.73 HD, which outperforms the best result of Swin-Unet on the Synapse dataset. CONCLUSIONS: The above improvements require only 4 h of training on a single NVIDIA Tesla V100, which is computationally cheap. Swin MoCo provides new approaches to improve the performance of tasks on small datasets. The code is publicly available at https://github.com/Zian-Xu/Swin-MoCo.

An anatomy-aware framework for automatic segmentation of parotid tumor from multimodal MRI.

Magnetic Resonance Imaging (MRI) plays an important role in diagnosing the parotid tumor, where accurate segmentation of tumors is highly desired for determining appropriate treatment plans and avoiding unnecessary surgery. However, the task remains nontrivial and challenging due to ambiguous boundaries and various sizes of the tumor, as well as the presence of a large number of anatomical structures around the parotid gland that are similar to the tumor. To overcome these problems, we propose a novel anatomy-aware framework for automatic segmentation of parotid tumors from multimodal MRI. First, a Transformer-based multimodal fusion network PT-Net is proposed in this paper. The encoder of PT-Net extracts and fuses contextual information from three modalities of MRI from coarse to fine, to obtain cross-modality and multi-scale tumor information. The decoder stacks the feature maps of different modalities and calibrates the multimodal information using the channel attention mechanism. Second, considering that the segmentation model is prone to be disturbed by similar anatomical structures and make wrong predictions, we design anatomy-aware loss. By calculating the distance between the activation regions of the prediction segmentation and the ground truth, our loss function forces the model to distinguish similar anatomical structures with the tumor and make correct predictions. Extensive experiments with MRI scans of the parotid tumor showed that our PT-Net achieved higher segmentation accuracy than existing networks. The anatomy-aware loss outperformed state-of-the-art loss functions for parotid tumor segmentation. Our framework can potentially improve the quality of preoperative diagnosis and surgery planning of parotid tumors.

Swin MAE: Masked autoencoders for small datasets.

The development of deep learning models in medical image analysis is majorly limited by the lack of large-sized and well-annotated datasets. Unsupervised learning does not require labels and is more suitable for solving medical image analysis problems. However, most unsupervised learning methods must be applied to large datasets. To make unsupervised learning applicable to small datasets, we proposed Swin MAE, a masked autoencoder with Swin Transformer as its backbone. Even on a dataset of only a few thousand medical images, Swin MAE can still learn useful semantic features purely from images without using any pre-trained models. It can equal or even slightly outperform the supervised model obtained by Swin Transformer trained on ImageNet in the transfer learning results of downstream tasks. Compared to MAE, Swin MAE brought a performance improvement of twice and five times for downstream tasks on BTCV and our parotid dataset, respectively. The code is publicly available at https://github.com/Zian-Xu/Swin-MAE.

Automated On-Line Microdialysis Sampling Coupled with HPLC for Synchronous Determination of Puerarin in Subcutaneous Tissue and Plasma Following Topical Administration.

BACKGROUND: Studies on transdermal administration have shown that puerarin can permeate rat skin rapidly with long-term drug delivery, but there are no reports demonstrating whether topical use of puerarin can provide a steady plasma concentration to produce therapeutic effects. The aim of the study is to evaluate the percutaneous penetration and plasma concentration of puerarin after transdermal administration in experimental rats. METHODS: The skin and plasma concentration of puerarin was quantified by microdialysis, and the recovery was determined by retrodialysis. Puerarin microdialysate concentrations were measured by on-line high-performance liquid chromatography (HPLC). Puerarin release from gels was determined by analysis of the amount of remaining drug after dermal application to hairless skin. RESULTS: The average recoveries of puerarin in the skin and plasma over an 8-hour period were 31.49% and 15.5%. Puerarin was rapidly absorbed with transdermal administration, with the C(max) values of 30.64 µg/mL and 3.53 µg/mL, the AUC0 t-values of 11.60 and 1.48 µg/mL per minute, for skin and plasma, respectively. CONCLUSIONS: The results indicate that the automated on-line microdialysis technique can be used to detect the skin and plasma pharmacokinetics of puerarin and that the use of skin gel can provide an effective means of puerarin administration.

[Pharmacokinetics of scopolamine hydrobromide oral disintegrative microencapsule tablets in Beagle dogs determined with LC-MS/MS].

The study aims to elucidate the characteristics of pharmacokinetics of scopolamine hydrobromide oral disintegrative microencapsule tablets in healthy Beagle dogs. Chromatographic separation was performed on a C18 column (100 mm x 3.0 mm, 3.5 microm) with methanol - 2 mmol x L(-1) ammonium formate (25 : 75) as the mobile phase. A trip-quadrupole tandem mass spectrum with the electrospray ionization (ESI) source was applied and positive ion multiple reaction monitoring mode was operated. Six Beagle dogs were randomly devided into two groups. They received oral single dose of scopolamine hydrobromide oral disintegrative microencapsule tablets 0.6 mg (test tablet) or scopolamine hydrobromide normal tablets (reference tablet). Plasma samples were collected at designed time. Plasma concentration of scopolamine hydrobromide was determined by LC-MS/MS and pharmacokinetic parameters were calculated. The pharmacokinetic parameters of test tablet vs reference tablet were as follows: C(max): (8.16 +/- 0.67) ng x mL(-1) vs (3.54 +/- 0.64) ng x mL(-1); t1/2: (2.83 +/- 0.45) h vs (3.85 +/- 0.82) h; t(max): (1.25 +/- 0.27) h vs (0.42 +/- 0.09) h; AUC(0-12h): (25.06 +/- 3.75) h x ng x mL(-1) vs (9.59 +/- 1.02) h x ng x mL(-1); AUC(0-infinity): (26.30 +/- 3.92) h x ng x mL(-1) vs (10.80 +/- 1.45) h x ng x mL(-1); MRT(0-12h): (3.38 +/- 0.34) h vs (3.86 +/- 0.26) h; MRT(0-infinity): (3.98 +/- 0.63) h vs (5.37 +/- 1.00) h. The absorption rate and AUC of test tablet is different from that of reference tablet. The bioavailability of test tablet is better than those of reference tablet.

[Enzyme kinetics of ligustilide metabolism in rat liver microsomes].

To study the enzyme kinetics of ligustilide metabolism and the effects of selective CYP450 inhibitors on the metabolism of ligustilide in liver microsomes of rat, a LC-MS method was established for quantitative analysis of ligustilide in liver microsomes incubation system with nitrendipine as internal standard. The determination m/z for ligustilide was 173, and for nitrendipine, 315. An optimum incubation system was found and various selective CYP inhibitors were used to investigate their inhibitory effects on the metabolism of ligustilide. The results showed that enzyme kinetics of ligustilide could be significantly inhibited by ketoconazole, trimethoprim and a-naphthoflavon but scarcely inhibited by omeprazole, 4-methylpyrazole and quinidine. Therefore, CYP3A4, CYP2C9 and CYP1A2 are the major isoenzyme participated in in vitro metabolism of ligustilide.

Determination of oxymatrine and its metabolite matrine in rat blood and dermal microdialysates by high throughput liquid chromatography/tandem mass spectrometry.

Oxymatrine (OMT) and matrine (MT) are the major quinolizidine alkaloids found in certain Sophora plants, which have been extensively used in China for the treatment of viral hepatitis, cancer, cardiac diseases and skin diseases (such as atopic dermatitis and eczema). A precise, sensitive and high throughput LC-MS/MS was developed to determine OMT and its metabolite MT in rat blood and dermis collected using microdialysis technique. Microdialysis probes were inserted into the jugular vein/right atrium and dermis of Wistar rats, and 3% OMT gel (1g) was administered via topical application. The samples were collected and then injected into the LC-MS/MS system after adding the internal standard (codeine, CDN). Chromatographic separation was achieved in a run time of 2min on a reversed phase short-column (50mmx2.1mm, 3.5microm). The mobile phase for column separation was methanol-ammonium formate (pH 5.0; 25mM) (70:30, v/v) with a flow rate of 0.3mL/min. A diverter valve was installed post-LC column for desalting. Detection of analytes and IS was done by tandem mass spectrometry, operating in positive ion and multiple reaction monitoring (MRM) acquisition mode. The protonated precursor to product ion transitions monitored for OMT, MT and IS was m/z 265.0-->247.3, 249.1-->148.3 and 300.0-->215.2, respectively. The lower limit of quantification (LLOQ) for OMT and MT was 0.5ng/mL. The calibration curves were linear over the range of 0.5-1000ng/mL for OMT and MT with a coefficient of determination >0.999. This selective and sensitive method is useful for the determination of OMT and MT and in the pharmacokinetic studies of these compounds. The blood and dermal concentration-time profile of OMT and its metabolite MT suggest that the limiting factor for dermal metabolism is the low capacity of enzymes in the skin rather than the quantity of penetrated OMT.

Assessment of in vitro and in vivo recovery of sinomenine using microdialysis.

BACKGROUND/PURPOSE: For microdialysis studies in the skin, laboratory-made linear probes are often used. The application of the microdialysis technique to the investigation of pharmacokinetics and pharmacodynamics of drugs requires careful assessment of the linear probes' performance to ensure validity of the data obtained using this technique. The aim of this study was to establish and validate the microdialysis technique for investigation of the pharmacokinetics and pharmacodynamics of sinomenine. METHODS: Using different lengths of the dialysis membrance and different perfusion flow rates, a flow rate of 2 microL/min with 20-min sampling intervals was selected for the subsequent studies, based on the analysis of sinomenine from the microdialysis probe. In vitro recovery of sinomenine from the microdialysis probe was independent of concentration, stable over an 8-h period. Comparable in vitro recoveries were obtained by different established approaches including recovery estimation by gain, loss and the zero-net flux (ZNF) method. Recovery by loss was used to study the in vivo recovery of sinomenine from rat subcutaneous tissue. RESULTS: The performance of the microdialysis system was stable over an 8-h study, resulting in a mean in vitro recovery of 51.91+/-1.29%. Recovery obtained using the ZNF plot was 52.43%. In vivo recovery of sinomenine was 34.46+/-0.76% and was stable over the 7-h study period. CONCLUSION: The in vitro and in vivo performance of the microdialysis technique was established for the study of sinomenine. It would prove to be a useful and reliable tool to study the pharmacokinetics and pharmacodynamics of sinomenine. The data obtained in our study highlight the importance of a systematic examination of microdialysis linear probe validation.

Pharmacokinetic study of free-form sinomenine in rat skin by microdialysis coupled with liquid chromatography-electrospray mass spectrometry.

Sinomenine (7,8-didehydro-4-hydroxy-3,7-dimethoxy-17-methylmorphinan-6-one) is a pure alkaloid extracted from the Chinese medical plant. In this report a liquid chromatography-electrospray mass spectrometry (LC-ESI-MS) method with in vivo microdialysis for the pharmacokinetic study of free-form sinomenine in rat skin has been developed. A microdialysis probe was surgically implanted into the subcutaneous tissue of the rats and an isotonic phosphate buffer (PBS) was used as the perfusion medium. Samples were collected and then analyzed off-line by LC-ESI-MS. The chromatographic separation was achieved within 4.2 min by using a narrow-bore Xterra C(18) column (2.1 x 150 mm, 5 microm) with acetonitrile-(10 mmol/L ammonium acetate buffer, 0.1% acetic acid) (15:85, v/v). Ion signal m/z 330.1 for sinomenine was measured in the positive mode. Linearity was established for the range of concentrations of 2.0-10000.0 ng/mL with a coefficient of determination (r) of 0.9989. The intra- and inter-day reproducibility of the present method was better than 6%. The lower limit of quantification (LLOQ) was 1.0 ng/mL. The proposed method described provides more authentic information on pharmacokinetics and metabolism at the site of action by using the coupling of microdialysis to LC-ESI-MS technique than the traditional sampling methods.