Unsupervised Segmentation of 5D Hyperpolarized Carbon-13 MRI Data Using a Fuzzy Markov Random Field Model

Taxonomy of hybrid architectures involving rule-based reasoning and machine learning in clinical decision systems: A scoping review

BACKGROUND

As the application of Artificial Intelligence (AI) technologies increases in the healthcare sector, the industry faces a need to combine medical knowledge, often expressed as clinical rules, with advances in machine learning (ML), which offer high prediction accuracy at the expense of transparency of decision making.

PURPOSE

This paper seeks to review the present literature, identify hybrid architecture patterns that incorporate rules and machine learning, and evaluate the rationale behind their selection to inform future development and research on the design of transparent and precise clinical decision systems.

METHODS

PubMed, IEEE Explore, and Google Scholar were queried in search for papers from 1992 to 2022, with the keywords: "clinical decision system", "hybrid clinical architecture", "machine learning and clinical rules". Excluded articles did not use both ML and rules or did not provide any explanation of employed architecture. A proposed taxonomy was used to organize the results, analyze them, and depict them in graphical and tabular form. Two researchers, one with expertise in rule-based systems and another in ML, reviewed identified papers and discussed the work to minimize bias, and the third one re-reviewed the work to ensure consistency of reporting.

RESULTS

The authors screened 957 papers and reviewed 71 that met their criteria. Five distinct architecture archetypes were determined: Rules are Embedded in ML architecture (REML) (most used), ML pre-processes input data for Rule-Based inference (MLRB), Rule-Based method pre-processes input data for ML prediction (RBML), Rules influence ML training (RMLT), Parallel Ensemble of Rules and ML (PERML), which was rarely observed in clinical contexts.

CONCLUSIONS

Most architectures in the reviewed literature prioritize prediction accuracy over explainability and trustworthiness, which has led to more complex embedded approaches. Alternatively, parallel (PERML) architectures may be employed, allowing for a more transparent system that is easier to explain to patients and clinicians. The potential of this approach warrants further research.

OTHER

A limitation of the study may be that it reviews scientific literature, while algorithms implemented in clinical practice may present different distributions of motivations and implementations of hybrid architectures.

Hexagonal-Grid-Layout Image Segmentation Using Shock Filters: Computational Complexity Case Study for Microarray Image Analysis Related to Machine Learning Approaches

Hexagonal grid layouts are advantageous in microarray technology; however, hexagonal grids appear in many fields, especially given the rise of new nanostructures and metamaterials, leading to the need for image analysis on such structures. This work proposes a shock-filter-based approach driven by mathematical morphology for the segmentation of image objects disposed in a hexagonal grid. The original image is decomposed into a pair of rectangular grids, such that their superposition generates the initial image. Within each rectangular grid, the shock-filters are once again used to confine the foreground information for each image object into an area of interest. The proposed methodology was successfully applied for microarray spot segmentation, whereas its character of generality is underlined by the segmentation results obtained for two other types of hexagonal grid layouts. Considering the segmentation accuracy through specific quality measures for microarray images, such as the mean absolute error and the coefficient of variation, high correlations of our computed spot intensity features with the annotated reference values were found, indicating the reliability of the proposed approach. Moreover, taking into account that the shock-filter PDE formalism is targeting the one-dimensional luminance profile function, the computational complexity to determine the grid is minimized. The order of growth for the computational complexity of our approach is at least one order of magnitude lower when compared with state-of-the-art microarray segmentation approaches, ranging from classical to machine learning ones.

The potential of hyperpolarised 13C-MRI to target glycolytic tumour core in prostate cancer

Hyperpolarised [1-13C]pyruvate MRI (HP-13C-MRI) is an emerging metabolic imaging technique that has shown promise for evaluating prostate cancer (PCa) aggressiveness. Accurate tumour delineation on HP-13C-MRI is vital for quantitative assessment of the underlying tissue metabolism. However, there is no consensus on the optimum method for segmenting HP-13C-MRI, and whole-mount pathology (WMP) as the histopathological gold-standard is only available for surgical patients. Although proton MRI can be used for tumour delineation, this approach significantly underestimates tumour volume, and metabolic tumour segmentation based on HP-13C-MRI could provide an important functional metric of tumour volume. In this study, we quantified metabolism using HP-13C-MRI and segmentation approaches based on WMP maps, 1H-MRI-derived T2-weighted imaging (T2WI), and HP-13C-MRI-derived total carbon signal-to-noise ratio maps (TC-SNR) with an SNR threshold of 5.0. 13C-labelled pyruvate SNR, lactate SNR, TC-SNR, and the pyruvate-to-lactate exchange rate constant (kPL) were significantly higher when measured using the TC-SNR-guided approach, which also corresponded to a significantly higher tumour epithelial expression on RNAscope imaging of the enzyme catalysing pyruvate-to-lactate metabolism (lactate dehydrogenase (LDH)). However, linear regression and Bland-Altman analyses demonstrated a strong linear relationship between all three segmentation approaches, which correlated significantly with RNA-scope-derived epithelial LDH expression. These results suggest that standard-of-care T2WI and TC-SNR maps could be used as clinical reference tools for segmenting localised PCa on HP-13C-MRI in the absence of the WMP gold standard. The TC-SNR-guided approach could be used clinically to target biopsies towards highly glycolytic tumour areas and therefore to sample aggressive disease with higher precision. KEY POINTS: • T2WI- and TC-SNR-guided segmentations can be used in all PCa patients and do not explicitly require WMP maps. • Agreement between the three segmentation approaches is biologically validated by their strong relationship with epithelial LDH mRNA expression. • The TC-SNR-guided approach can potentially be used to identify occult disease on 1H-MRI and target the most glycolytically active regions.

Enhancing the spatial resolution of hyperpolarized carbon-13 MRI of human brain metabolism using structure guidance

PURPOSE

Dynamic nuclear polarization is an emerging imaging method that allows noninvasive investigation of tissue metabolism. However, the relatively low metabolic spatial resolution that can be achieved limits some applications, and improving this resolution could have important implications for the technique.

METHODS

We propose to enhance the 3D resolution of carbon-13 magnetic resonance imaging (13 C-MRI) using the structural information provided by hydrogen-1 MRI (1 H-MRI). The proposed approach relies on variational regularization in 3D with a directional total variation regularizer, resulting in a convex optimization problem which is robust with respect to the parameters and can efficiently be solved by many standard optimization algorithms. Validation was carried out using an in silico phantom, an in vitro phantom and in vivo data from four human volunteers.

RESULTS

The clinical data used in this study were upsampled by a factor of 4 in-plane and by a factor of 15 out-of-plane, thereby revealing occult information. A key finding is that 3D super-resolution shows superior performance compared to several 2D super-resolution approaches: for example, for the in silico data, the mean-squared-error was reduced by around 40% and for all data produced increased anatomical definition of the metabolic imaging.

CONCLUSION

The proposed approach generates images with enhanced anatomical resolution while largely preserving the quantitative measurements of metabolism. Although the work requires clinical validation against tissue measures of metabolism, it offers great potential in the field of 13 C-MRI and could significantly improve image quality in the future.

B-Map: a fuzzy-based model to detect foreign objects in a brain

To cope up with the medical complications, scientists and physicians rely more on digitized historical evidence. It helps them to identify the disease and to develop new drugs and strategies. The authors have designed a model called B-Map. It can detect and segmenting any foreign object in the brain using fuzzy rules. The model can detect objects such as cancer and brain tumor. The proposed work aims at designing a classifier. The classifier would help to detect all possible foreign objects using one application. B-Map has been compared with benchmark algorithms such as K-means and ANN. It was found that the proposed model performs significantly better than the current techniques. Original patients' sample reports are taken from various medical laboratories. The figure numbers are retained as in the paper. The proposed model is able to find the edges and segment different types of foreign objects or one can say unexpected developments. Figure 12 shows the outer edges of a section of a brain MRI. The patient's MRI very clearly shows Hydrocephalus. The same is segmented and shown in Fig. 13. Figure 14 shows a segment of benign development and 15 shows a cancerous development which are again successfully segmented by the proposed model. The data on which testing is done is clinical data of the original patients. As the patient's details and data cannot be shared the author's cannot upload the data in the repository. As soon as the research completes, a benchmark dataset will be created and uploaded in public domain so that researchers can access it.

Quantifying normal human brain metabolism using hyperpolarized [1-13C]pyruvate and magnetic resonance imaging

Hyperpolarized 13C Magnetic Resonance Imaging (13C-MRI) provides a highly sensitive tool to probe tissue metabolism in vivo and has recently been translated into clinical studies. We report the cerebral metabolism of intravenously injected hyperpolarized [1-13C]pyruvate in the brain of healthy human volunteers for the first time. Dynamic acquisition of 13C images demonstrated 13C-labeling of both lactate and bicarbonate, catalyzed by cytosolic lactate dehydrogenase and mitochondrial pyruvate dehydrogenase respectively. This demonstrates that both enzymes can be probed in vivo in the presence of an intact blood-brain barrier: the measured apparent exchange rate constant (kPL) for exchange of the hyperpolarized 13C label between [1-13C]pyruvate and the endogenous lactate pool was 0.012 ± 0.006 s-1 and the apparent rate constant (kPB) for the irreversible flux of [1-13C]pyruvate to [13C]bicarbonate was 0.002 ± 0.002 s-1. Imaging also revealed that [1-13C]pyruvate, [1-13C]lactate and [13C]bicarbonate were significantly higher in gray matter compared to white matter. Imaging normal brain metabolism with hyperpolarized [1-13C]pyruvate and subsequent quantification, have important implications for interpreting pathological cerebral metabolism in future studies.

Hyperpolarized carbon-13 magnetic resonance spectroscopic imaging: a clinical tool for studying tumour metabolism

Glucose metabolism in tumours is reprogrammed away from oxidative metabolism, even in the presence of oxygen. Non-invasive imaging techniques can probe these alterations in cancer metabolism providing tools to detect tumours and their response to therapy. Although Positron Emission Tomography with (18F)2-fluoro-2-deoxy-D-glucose (18F-FDG PET) is an established clinical tool to probe cancer metabolism, it has poor spatial resolution and soft tissue contrast, utilizes ionizing radiation and only probes glucose uptake and phosphorylation and not further downstream metabolism. Magnetic Resonance Spectroscopy (MRS) has the capability to non-invasively detect and distinguish molecules within tissue but has low sensitivity and can only detect selected nuclei. Dynamic Nuclear Polarization (DNP) is a technique which greatly increases the signal-to-noise ratio (SNR) achieved with MR by significantly increasing nuclear spin polarization and this method has now been translated into human imaging. This review provides a brief overview of this process, also termed Hyperpolarized Carbon-13 Magnetic Resonance Spectroscopic Imaging (HP 13C-MRSI), its applications in preclinical imaging, an outline of the current human trials that are ongoing, as well as future potential applications in oncology.