Applications of Artificial Intelligence in Pediatric Oncology: A Systematic Review

PURPOSE

There is a need for an improved understanding of clinical and biologic risk factors in pediatric cancer to improve patient outcomes. Machine learning (ML) represents the application of computational inference from advanced statistical methods that can be applied to increasing amount of data available for study in pediatric oncology. The goal of this systematic review was to systematically characterize the state of ML in pediatric oncology and highlight advances and opportunities in the field.

METHODS

We conducted a systematic review of the Embase, Scopus, and MEDLINE databases for applications of ML in pediatric oncology. Query results from all three databases were aggregated and duplicate studies were removed.

RESULTS

A total of 42 unique articles that examined the applications of ML in pediatric oncology met inclusion criteria for review. We identified 20 studies of CNS tumors, 13 of solid tumors, and nine of leukemia. ML tasks included classification, prediction of treatment response, and dose optimization with a variety of methods being used including neural network, k-nearest neighbor, random forest, naive Bayes, and support vector machines. Strengths of the identified studies included matching or outperforming physician comparators via automated analysis and predicting therapeutic response. Common limitations included significant heterogeneity in reporting standards, clinical applicability, small sample sizes, and missing external validation cohorts.

CONCLUSION

We identified areas where ML can enhance clinical care in ways that may not otherwise be achievable. Although ML promises enormous potential in improving diagnostics, decision making, and monitoring for children with cancer, the field remains in early stages and future work will be aided by standards and guidelines to ensure rigorous methodologic design and maximizing clinical utility.

Assessment of natural DNA methylation variation and its association with economically important traits in dolichos bean (Lablab purpureus L. Var. Lignosus) using AMP-PCR assay

As a prelude to exploit DNA methylation-induced variation, we hypothesized the existence of substantial natural DNA methylation variation and its association with economically important traits in dolichos bean, and tested it using amplified methylation polymorphism-polymerase chain reaction (AMP-PCR) assay. DNA methylation patterns such as internal, external, full and non-methylation were amplified in a set of 64 genotypes using 26 customized randomly amplified polymorphic DNA (RAPD) primers containing 5'CCGG3' sequence. The 64 genotypes included 60 germplasm accessions (GA), two advanced breeding lines (ABLs) and two released varieties. The ABLs and released varieties are referred to as improved germplasm accessions (IGA) in this study. The association of DNA methylation patterns with economically important traits such as days to 50% flowering, raceme length, fresh pods plant^-1, fresh pod yield plant^-1 and 100-fresh seed weight was explored. At least 50 genotypes were polymorphic for DNA methylation patterns at 10 loci generated by seven of the 26 RAPD primers. The GA and IGA differed significantly for total, full and external methylation and the frequency of methylation was higher in GA compared to that in IGA. The genotypes with external methylation produced longer racemes than those with full, internal and non-methylation in that order at polymorphic RAPD-11-242 locus. High pod yielding genotypes had significantly lower frequency of full methylation than low yielding ones. On the contrary, the genotypes that produced heavier fresh seeds harboured higher frequencies of total and externally methylated loci than those that produced lighter fresh seeds.

Predicting Response to Chemotherapy in Patients With Newly Diagnosed High-Risk Neuroblastoma: A Report From the International Neuroblastoma Risk Group

PURPOSE

Metaiodobenzylguanidine (MIBG) scans are a radionucleotide imaging modality that undergo Curie scoring to semiquantitatively assess neuroblastoma burden, which can be used as a marker of therapy response. We hypothesized that a convolutional neural network (CNN) could be developed that uses diagnostic MIBG scans to predict response to induction chemotherapy.

METHODS

We analyzed MIBG scans housed in the International Neuroblastoma Risk Group Data Commons from patients enrolled in the Children's Oncology Group high-risk neuroblastoma study ANBL12P1. The primary outcome was response to upfront chemotherapy, defined as a Curie score ≤ 2 after four cycles of induction chemotherapy. We derived and validated a CNN using two-dimensional whole-body MIBG scans from diagnosis and evaluated model performance using area under the receiver operating characteristic curve (AUC). We also developed a clinical classification model to predict response on the basis of age, stage, and MYCN amplification.

RESULTS

Among 103 patients with high-risk neuroblastoma included in the final cohort, 67 (65%) were responders. Performance in predicting response to upfront chemotherapy was equivalent using the CNN and the clinical model. Class-activation heatmaps verified that the CNN used areas of disease within the MIBG scans to make predictions. Furthermore, integrating predictions using a geometric mean approach improved detection of responders to upfront chemotherapy (geometric mean AUC 0.73 v CNN AUC 0.63, P < .05; v clinical model AUC 0.65, P < .05).

CONCLUSION

We demonstrate feasibility in using machine learning of diagnostic MIBG scans to predict response to induction chemotherapy for patients with high-risk neuroblastoma. We highlight improvements when clinical risk factors are also integrated, laying the foundation for using a multimodal approach to guiding treatment decisions for patients with high-risk neuroblastoma.

Discovery of genomic regions associated with resistance to late wilt disease caused by Harpophora maydis (Samra, Sabet and Hing) in maize (Zea mays L.)

Late wilt disease (LWD) caused by Harpophora maydis (Samra, Sabet and Hing) is emerging as major production constraint in maize across the world. As a prelude to develop maize hybrid resistance to LWD, genetic basis of resistance was investigated. Two F_2:3 mapping populations (derived from CV156670 × 414-33 (P-1) and CV156670 × CV143587 (P-2)) were challenged with LWD at two locations (Kallinayakanahalli and Muppadighatta) during 2017 post-rainy season. A wider range of LWD scores was observed at both locations in both the populations. LWD response was influenced by significant genotype × location interaction. Six and 56 F_2:3 progeny families showed resistance level better than resistant parent. A total of 150 and 199 polymorphic single nucleotide polymorphism markers were used to genotype P-1 and P-2, respectively. Inclusive composite interval mapping was performed to detect significant Quantitative Trait Loci (QTL), QTL × QTL, QTL × location interaction effects. Three major and four minor QTL controlling LWD resistance were detected on chromosome-1. The position and effect of the QTL varied with the location. Significant di-QTL interactions involving QTL (with significant and/or non-significant effects) located within and between all the ten chromosomes were detected. Five of the seven detected QTL showed significant QTL × location interaction. Though two major QTL (q-lw-1.5 and q-lw-1.6) with lower Q×L interaction effects could be considered as stable, their phenotypic variance is not large enough to deploy them in Marker Assisted Selection (MAS). However, these QTL are of paramount importance in accumulating positive alleles for LWD resistance breeding.

Long-term outcomes in children with Hirschsprung's disease and transition zone bowel pull-through: impact of surgical techniques and role for conservative approach

PURPOSE

Presence of transition zone (TZ) in the pulled colon can impact the outcome of surgery in children with Hirschsprung's disease. There is a wide variation in terminology used to define TZ and its management. We present our series of managing 11 such children with considerations for conservative management.

METHODS

Eleven of 114 children operated for Hirschsprung's disease had features of TZ on the 4-quadrant doughnut assessment of proximal anastomosing margin. They were followed up for development of obstructive symptoms, failure of pull-through procedure or bowel-related complications. Intervention done were observation with laxatives, dilatation, Botox injection and redo pull-through.

RESULTS

Of the 11 children, 6 underwent Duhamel's procedure and 5, transanal endorectal pull-through (TERP). Features identified on HPE were presence of hypertrophic nerve bundles involving 2 or 3 quadrants in the circumferential doughnut biopsy of proximal anastomosing margin. Observed symptoms included constipation, enterocolitis, increased bowel frequency and soiling. Intervention done were use of laxatives with bowel management program in six and Botox injections in four. Only one child with TZ in 3 quadrants required redo surgery. Mean follow-up was 5.2 years with resolution of symptoms in most.

CONCLUSION

This study highlights the role of conservative management with good outcomes in children with TZ bowel pull-through having hypertrophic nerve fibers and normal ganglion pattern. Children who underwent Duhamel's procedure had little impact with the presence of TZ at anastomotic margin and majority of those undergoing TERP benefitted from Botox injection. Conservative management can be attempted successfully to prevent redo surgical interventions as they can lead to poorer outcomes. Only those children not responding to conservative measures need to be planned for revision surgery.

Predicting response to chemotherapy in neuroblastoma using deep learning: A report from the International Neuroblastoma Risk Group

10039

Background: Metaiodobenzylguanidine (MIBG) scans are a radionucleotide imaging modality used to evaluate neuroblastoma stage at diagnosis and also determine disease response following therapy. Curie scoring is used to semi-quantitatively assess disease burden from an MIBG scan on a scale from none (0) to widespread throughout the body (30). While a Curie score ≤2 after six cycles of induction chemotherapy has been shown to be prognostic of outcome, there is no established correlation between diagnostic Curie score and outcome. Deep learning models, such as convolutional neural networks (CNN), have been shown to learn generalizable patterns within images for successful classification of metastases and detection of multiple adult cancers. We hypothesized a CNN could be developed to predict response to induction chemotherapy, a proxy for outcome, using diagnostic MIBG scans. Methods: DICOM MIBG scans and associated clinical data from a Children’s Oncology Group (COG) pilot study for children diagnosed with high-risk neuroblastoma (ANBL12P1; NCT1798004) were deidentified and linked to clinical data by the Pediatric Cancer Data Commons and obtained from the International Neuroblastoma Risk Group Data Commons. Patients were defined as having a poor response to induction chemotherapy if their Curie score after four cycles of induction chemotherapy was ≥2. An independent external validation cohort was comprised of 29 images from 26 high-risk patients treated at the University of Chicago with clinically-annotated diagnostic and post-cycle six induction DICOM MIBG scans. The CNN was trained using 2D whole body MIBG scans obtained at diagnosis. We developed the CNN using a transfer learning approach using the Xception architecture as the base layer. Hyperparameter optimization was performed using an 80%-20% train-validation strategy. Model performance was evaluated using area under the receiver operating characteristic curve (AUROC). Results: Among 146 patients with high-risk neuroblastoma enrolled on ANBL12P1, 104 had available diagnostic and end-induction MIBG scans. There were no differences in clinical or biological characteristics between included and excluded patients. The base model CNN was able to predict which patients had a poor response to induction chemotherapy with an AUROC of 0.72 in the validation set from the ANBL12P1 cohort. Additionally, the CNN was able to predict patient response to therapy with an AUROC of 0.64 in an independent external dataset from University of Chicago. Conclusions: Our study suggests it is feasible to apply machine learning of diagnostic MIBG scans to predict response to chemotherapy for high-risk neuroblastoma patients. Given these promising results, further work to improve AUROC and performance within larger datasets is ongoing.

Optical diagnosis of oral lichen planus: A clinical study on the use of autofluorescence spectroscopy combined with multivariate analysis

Oral lichen planus (OLP) is a chronic mucocutaneous inflammatory condition of stratified squamous epithelia. OLP is a potentially malignant condition in oral mucosa. Patients with OLP have an increased risk of developing squamous cell carcinoma. Therefore an early and accurate diagnosis is necessary to avoid further damage to the oral mucosa. Biopsy followed by histopathological examination is the gold standard for the diagnosis of oral mucosal lesions including OLP. But this invasive procedure is traumatic and time consuming with limited statistical confidence level. Autofluorescence spectroscopy (AFS) has recently emerged as a potential tool to evaluate the biochemical changes associated with oral cavity disorders. In this study, we used AFS to differentiate the oral cavity tissue of 20 OLP patients from that of 16 normal volunteers. Spectra from oral mucosa were acquired at 280, 320 and 410 nm excitation wavelengths which correspond to the excitation energy of major endogenous fluorophores. Normalized spectral data at 320 nm excitation showed significant increase in the intensity of collagen peak for OLP. Optical redox ratio and total hemoglobin concentration estimated from the spectral data revealed significant increase and decrease respectively in OLP and normal patients. Principal component analysis followed by linear discriminant analysis (PCA-LDA) provided sensitivity and specificity of 71 and 80%, 80 and 90%, and 72 and 75% respectively for 280, 320 and 410 nm excited spectral datasets. Meanwhile, partial least square discriminant analysis (PLS-DA) provided sensitivity and specificity of 69 and 77%, 78 and 91% and 73 and 78.13% respectively for 280, 320 and 410 nm excited spectral datasets. From the results, it is concluded that AFS is an efficient tool for the non invasive diagnosis of OLP, with 320 nm light identified as the best wavelength for excitation.

Image analysis and data processing for COVID-19

COVID-19 is a deadly disease caused by the severe acute respiratory syndrome coronavirus (SARS-CoV-2). It was first discovered by variations in the respirational and immune systems of a patient who died of a severe acute respiratory syndrome. The first country heavily affected by coronavirus was China. The first case was detected in Wuhan city, China. This virus spreads rapidly from person to person. Based on laboratory tests for coronavirus disease in humans, it is suspected that bats are the natural source of spread of large varieties of virus. The two major viruses, SARS-CoV and Middle East respiratory syndrome coronavirus, originated from the bat; it caused an unexpected disease outbreak in the 21st century throughout the world. Researchers and doctors have investigated COVID in cadavers. The virus was detected in lung, trachea/bronchus, stomach, small intestine, distal convoluted renal tubule, sweat gland, pancreas, adrenal gland, parathyroid, pituitary, cerebrum, and liver. However, it was not noted in bone marrow, heart, aorta, cerebellum, thyroid, testis, esophagus, spleen, lymph node, ovary, muscle, or uterus. This chapter briefly discusses image analysis and data processing used to accelerate COVID-19 detection and support the efforts of researchers and physician to help infected people and break the chain of disease from person to person.

Cobalt Oxide Nanograins and Silver Nanoparticles Decorated Fibrous Polyaniline Nanocomposite as Battery-Type Electrode for High Performance Supercapattery

In this study, silver (Ag) and cobalt oxide (Co₃O₄) decorated polyaniline (PANI) fibers were prepared by the combination of in-situ aniline oxidative polymerization and the hydrothermal methodology. The morphology of the prepared Ag/Co₃O₄@PANI ternary nanocomposite was studied by scanning electron microscopy and transmission electron microscopy, while the structural studies were carried out by X-ray diffraction and X-ray photoelectron spectroscopy. The morphological characterization revealed fibrous shaped PANI, coated with Ag and Co₃O₄ nanograins, while the structural studies revealed high purity, good crystallinity, and slight interactions among the constituents of the Ag/Co₃O₄@PANI ternary nanocomposite. The electrochemical performance studies revealed the enhanced performance of the Ag/Co₃O₄@PANI nanocomposite due to the synergistic/additional effect of Ag, Co₃O₄ and PANI compared to pure PANI and Co₃O₄@PANI. The addition of the Ag and Co₃O₄ provided an extended site for faradaic reactions leading to the high specific capacity. The Ag/Co₃O₄@PANI ternary nanocomposite exhibited an excellent specific capacity of 262.62 C g⁻¹ at a scan rate of 3 mV s⁻¹. The maximum energy and power density were found to be 14.01 Wh kg⁻¹ and 165.00 W kg⁻¹, respectively. The cyclic stability of supercapattery (Ag/Co₃O₄@PANI//activated carbon) consisting of a battery type electrode demonstrated a gradual increase in specific capacity with a continuous charge–discharge cycle until ~1000 cycles, then remained stable until 2500 cycles and later started decreasing, thereby showing the cyclic stability of 121.03% of its initial value after 3500 cycles.

Fundamental Concepts of Hydrogels: Synthesis, Properties, and Their Applications

In the present review, we focused on the fundamental concepts of hydrogels-classification, the polymers involved, synthesis methods, types of hydrogels, properties, and applications of the hydrogel. Hydrogels can be synthesized from natural polymers, synthetic polymers, polymerizable synthetic monomers, and a combination of natural and synthetic polymers. Synthesis of hydrogels involves physical, chemical, and hybrid bonding. The bonding is formed via different routes, such as solution casting, solution mixing, bulk polymerization, free radical mechanism, radiation method, and interpenetrating network formation. The synthesized hydrogels have significant properties, such as mechanical strength, biocompatibility, biodegradability, swellability, and stimuli sensitivity. These properties are substantial for electrochemical and biomedical applications. Furthermore, this review emphasizes flexible and self-healable hydrogels as electrolytes for energy storage and energy conversion applications. Insufficient adhesiveness (less interfacial interaction) between electrodes and electrolytes and mechanical strength pose serious challenges, such as delamination of the supercapacitors, batteries, and solar cells. Owing to smart and aqueous hydrogels, robust mechanical strength, adhesiveness, stretchability, strain sensitivity, and self-healability are the critical factors that can identify the reliability and robustness of the energy storage and conversion devices. These devices are highly efficient and convenient for smart, light-weight, foldable electronics and modern pollution-free transportation in the current decade.