RBIO-03. INITIAL RESULT OF DEVELOP ROBUST DEEP LEARNING MODEL FOR DETECTING GENOMIC STATUS IN GLIOMAS AGAINST IMAGE DIFFERENCES AMONG FACILITIES

BACKGROUND The importance of detecting the genomic status of gliomas is increasingly recognized and IDH (isocitrate dehydrogenase) mutation and TERT (telomerase reverse transcriptase) promoter mutation have a significant impact on treatment decisions. Noninvasive prediction of these genomic statuses in gliomas is a challenging problem; however, a deep learning model using magnetic resonance imaging (MRI) can be a solution. The image differences among facilities causing performance degradation, called domain shift, have also been reported in other tasks such as brain tumor segmentation. We investigated whether a deep learning model could predict the gene status, and if so, to what extent it would be affected by domain shift. METHOD We used the Multimodal Brain Tumor Segmentation Challenge (BraTS) data and the Japanese cohort (JC) dataset consisted of brain tumor images collected from 544 patients in 10 facilities in Japan. We focused on IDH mutation and TERT promoter mutation. The deep learning models to predict the statuses of these genes were trained by the BraTS dataset or the training portion of the JC dataset, and the test portion of the JC dataset evaluated the accuracy of the models. RESULTS The IDH mutation predicting model trained by the BraTS dataset showed 80.0% accuracy for the validation portion of the BraTS dataset; however, only 67.3% for the test portion of the JC dataset. The TERT promoter mutation predicting model trained by the training portion of the JC dataset showed only 49% accuracy for the test portion of the JC dataset. CONCLUSION IDH mutation can be predicted by deep learning models using MRI, but the performance degeneration by domain shift was significant. On the other hand, TERT promoter mutation could not be predicted accurately enough by current deep learning techniques. In both mutations, further studies are needed.

Kilogram Sample Analysis by Nuclear Analytical Techniques: Complementary Opportunities for the Mineral and Geosciences

Sample-size reduction including homogenization is often required to obtain a test portion for element compositional analysis. Analyses of replicate test portions may provide insight into the sampling constant, and often much larger quantities are needed to limit the contribution of sampling error. In addition, it cannot be demonstrated that the finally obtained test portion is truly representative of the originally collected material. Nuclear analytical techniques such as neutron and photon activation analysis and (neutron-induced) prompt gamma activation analyses can now be used to study and overcome these analytical problems. These techniques are capable of obtaining multi-element measurements from irregularly shaped objects with masses ranging from multiple grams to multiple kilograms. Prompt gamma analysis can be combined with neutron tomography, resulting in position-sensitive information. The analysis of large samples provides unprecedented complementary opportunities for the mineral and geosciences. It enables the experimental assessment of the representativeness of test portions of the originally collected material, as well as the analysis of samples that are not allowed to be sub-sampled or dissolved, the analysis of materials that are difficult to be homogenized at large, and studies on the location of inhomogeneities. Examples of such applications of large-sample analyses are described herein.

Evaluation of the GENE-UP® EHEC Detection Method for the Detection of Enterohemorrhagic E. coli in Select Foods: Collaborative Study: First Action Method 2020.06

Background The GENE-UP® EHEC assay (Performance Tested MethodSM 121806) is a real-time PCR molecular detection method that utilizes Fluorescence Resonance Energy Transfer proprietary hybridization probes for the rapid detection of Enterohemorrhagic E. coli (EHEC) in select foods. Objective The purpose of this validation was to evaluate the method’s interlaboratory performance and submit the results to AOAC INTERNATIONAL for adoption as First Action Official Method of AnalysisSM for the detection of EHEC in select foods. Method The GENE-UP® method was evaluated in a multi-laboratory study as part of the MicroVal VALIDATION certification process using unpaired test portions for one food matrix, raw ground beef (85% lean). Collaborators evaluated the candidate method using either an automated or manual lysis procedure. The candidate method was compared to the ISO/TS 13136:2012 method. Data from 17 participants from 15 laboratories throughout the European Union was evaluated. Three levels of contamination were evaluated: a non-inoculated control level (0 CFU/test portion), a low contamination level (∼1 CFU/test portion) and a high contamination level (∼10 CFU/test portion). Data from the study were analyzed according to the probability of detection (POD) statistical model. Results The dLPODC values with 95% confidence interval between the candidate and reference method results were; –0.01 (–0.04, 0.02), 0.23 (0.07, 0.39) and 0.06 (0.01, 0.12) for the non-inoculated, low and high contamination levels, respectively. Conclusion For the candidate method, values obtained for repeatability and reproducibility were similar to the reference method and indicated minimal variation between samples or between laboratories. No discrepant results (false positive or false negative) were observed for each contamination. A statistical difference was calculated between the candidate and reference method at the low and high inoculation levels, with the candidate method detecting a higher number of positive samples indicating a higher sensitivity than the reference method. No differences in the recovery of the target analyte were observed between the manual and automated lysis procedures. Highlights The GENE-UP EHEC Detection Method provides end users a rapid, easy-to-use workflow for the detection of EHEC in food matrices.

Fine-Tuning Approach for Segmentation of Gliomas in Brain Magnetic Resonance Images with a Machine Learning Method to Normalize Image Differences among Facilities

Machine learning models for automated magnetic resonance image segmentation may be useful in aiding glioma detection. However, the image differences among facilities cause performance degradation and impede detection. This study proposes a method to solve this issue. We used the data from the Multimodal Brain Tumor Image Segmentation Benchmark (BraTS) and the Japanese cohort (JC) datasets. Three models for tumor segmentation are developed. In our methodology, the BraTS and JC models are trained on the BraTS and JC datasets, respectively, whereas the fine-tuning models are developed from the BraTS model and fine-tuned using the JC dataset. Our results show that the Dice coefficient score of the JC model for the test portion of the JC dataset was 0.779 ± 0.137, whereas that of the BraTS model was lower (0.717 ± 0.207). The mean Dice coefficient score of the fine-tuning model was 0.769 ± 0.138. There was a significant difference between the BraTS and JC models (p < 0.0001) and the BraTS and fine-tuning models (p = 0.002); however, no significant difference between the JC and fine-tuning models (p = 0.673). As our fine-tuning method requires fewer than 20 cases, this method is useful even in a facility where the number of glioma cases is small.

Improved Sample Selection and Preparation Methods for Sampling Plans Used to Facilitate Rapid and Reliable Estimation of Aflatoxin in Chicken Feed

Aflatoxin B1 (AFB1), a toxic fungal metabolite associated with human and animal diseases, is a natural contaminant encountered in agricultural commodities, food and feed. Heterogeneity of AFB1 makes risk estimation a challenge. To overcome this, novel sample selection, preparation and extraction steps were designed for representative sampling of chicken feed. Accuracy, precision, limits of detection and quantification, linearity, robustness and ruggedness were used as performance criteria to validate this modification and Horwitz function for evaluating precision. A modified sampling protocol that ensured representativeness is documented, including sample selection, sampling tools, random procedures, minimum size of field-collected aggregate samples (primary sampling), procedures for mass reduction to 2 kg laboratory (secondary sampling), 25 g test portion (tertiary sampling) and 1.3 g analytical samples (quaternary sampling). The improved coning and quartering procedure described herein (for secondary and tertiary sampling) has acceptable precision, with a Horwitz ratio (HorRat = 0.3) suitable for splitting of 25 g feed aliquots from laboratory samples (tertiary sampling). The water slurring innovation (quaternary sampling) increased aflatoxin extraction efficiency to 95.1% through reduction of both bias (−4.95) and variability of recovery (1.2–1.4) and improved both intra-laboratory precision (HorRat = 1.2–1.5) and within-laboratory reproducibility (HorRat = 0.9–1.3). Optimal extraction conditions are documented. The improved procedure showed satisfactory performance, good field applicability and reduced sample analysis turnaround time.

Evaluation of the GENE-UP ® Listeria spp. Method for the Detection of Listeria Species in a Variety of Foods and Select Environmental Surfaces: Collaborative Study, First Action 2019.10

Background The GENE-UP®Listeria spp. 2 (LIS 2) assay (Performance Tested MethodSM 121803) is a real-time PCR molecular detection method for the rapid detection of Listeria species (Listeria monocytogenes, L. innocua, L. ivanovii, L. seeligeri and L. welshimeri) in a variety of foods and environmental surfaces. Objective The purpose of this validation was to evaluate the method’s interlaboratory performance and submit the results to AOAC INTERNATIONAL for adoption as First Action Official MethodSM for the detection of Listeria species in a variety of foods and select environmental surfaces. Method The GENE-UP® method was evaluated in a multi-laboratory study as part of the AFNOR NF VALIDATION certification process using unpaired test portions for one food matrix, full-cream goat milk cottage cheese (8.4% fat). The candidate method was compared to the ISO 11290-1/Amd.1 reference method. Sixteen participants from 15 laboratories throughout the European Union participated. Three levels of contamination were evaluated: a non-inoculated control level (0 CFU/test portion), a low contamination level (∼2 CFU/test portion) and a high contamination level (∼10 CFU/test portion). Data from that study were analyzed according to the Probability of Detection (POD) statistical model. Results The dLPODC values with 95% confidence interval between the candidate and reference method results were; -0.02 (-0.07, 0.03), -0.08 (-0.31, 0.16) and 0.00 (-0.03, 0.03) for the non-inoculated, low and high contamination levels respectively. Conclusion The dLPODC results demonstrate no difference in performance between the candidate method and reference method for the matrix evaluated. Highlights Data from a singular collaborative study was used to achieve adoption as AOAC First Action Official Method for the detection of Listeria species in a variety of foods and select environmental surfaces.

Characterisation and composition identification of waste-derived fuels obtained from municipal solid waste using thermogravimetry: A review

Thermogravimetric analysis (TGA) is the most widespread thermal analytical technique applied to waste materials. By way of critical review, we establish a theoretical framework for the use of TGA under non-isothermal conditions for compositional analysis of waste-derived fuels from municipal solid waste (MSW) (solid recovered fuel (SRF), or refuse-derived fuel (RDF)). Thermal behaviour of SRF/RDF is described as a complex mixture of several components at multiple levels (including an assembly of prevalent waste items, materials, and chemical compounds); and, operating conditions applied to TGA experiments of SRF/RDF are summarised. SRF/RDF mainly contains cellulose, hemicellulose, lignin, polyethylene, polypropylene, and polyethylene terephthalate. Polyvinyl chloride is also used in simulated samples, for its high chlorine content. We discuss the main limitations for TGA-based compositional analysis of SRF/RDF, due to inherently heterogeneous composition of MSW at multiple levels, overlapping degradation areas, and potential interaction effects among waste components and cross-contamination. Optimal generic TGA settings are highlighted (inert atmosphere and low heating rate (⩽10°C), sufficient temperature range for material degradation (⩾750°C), and representative amount of test portion). There is high potential to develop TGA-based composition identification and wider quality assurance and control methods using advanced thermo-analytical techniques (e.g. TGA with evolved gas analysis), coupled with statistical data analytics.

Evaluation of the GENE-UP ® Listeria monocytogenes Method for the Detection of Listeria monocytogenes in a Variety of Foods and Select Environmental Surfaces: Collaborative Study, First Action 2019.11

Background The GENE-UP®Listeria monocytogenes 2 (LMO 2) assay (Performance Tested MethodSM 121804) uses real-time PCR technology and a proprietary detection platform, the GENE-UP® Thermocycler, to detect Listeria monocytogenes in a variety of foods and environmental surfaces. Objective The purpose of this validation was to evaluate the method’s interlaboratory performance and submit the result to AOAC INTERNATIONAL for adoption as First Action Official MethodSM for the detection of Listeria monocytogenes in a variety of foods and select environmental surfaces. Method The GENE-UP® method was evaluated in a multi-laboratory study as part of the AFNOR NF VALIDATION certification process using unpaired test portions for one food matrix, full-cream goat milk cottage cheese (8.4% fat). The candidate method was compared to the ISO 11290-1/Amd.1:2004 reference method. Sixteen participants from 15 laboratories throughout the European Union participated. Three levels of contamination were evaluated: a non-inoculated control level (0 CFU/test portion), a low inoculum level (∼2 CFU/test portion) and a high inoculum level (∼10 CFU/test portion). Data from the study were analyzed according to the Probability of Detection (POD) statistical model as presented in the AOAC validation guidelines. Results The dLPODC values with 95% confidence interval for each comparison were; -0.02 (-0.07, 0.03), -0.08 (-0.31, 0.16) and 0.00 (-0.03, 0.03) for the non-inoculated, low and high contamination levels respectively. Conclusion The dLPODC results demonstrate no difference in performance between the candidate method and reference method for the matrix evaluated. Highlights The GENE-UP LMO method demonstrated accuracy and precision in detecting and discerning L. monocytogenes from other Listeria species.

A Study to Evaluate the Need for Comminution in the Preparation of Food Materials for Testing of Listeria monocytogenes, Staphylococcus Species, and Generic Escherichia coli

Background Comminution reduces the sampling error arising from distributional heterogeneity of the target contaminant/target analyte in the material, facilitating the selection of a more representative test portion. A laboratory sampling method incorporating comminution prior to selection of the test portion (Sampling Method B) was compared to current sampling methods that used no comminution step (Sampling Method A). Objective This required the development of an efficient process for comminution of food samples prior to removal of the test portion for the detection and isolation of Listeria monocytogenes and the enumeration of Staphylococcus species and Escherichia coli. Method From December 2016 to December 2017, 2742 tests were conducted on 778 unique food samples. For all food samples, a test portion (TPA) was first removed using Sampling Method A, and then the remainder of the material was comminuted and a second test portion (TPB) was removed using Sampling Method B and tested alongside the first portion. Results Across all food matrices and microbial targets, 17 additional targets were detected using only Sampling Method B, and positive detections of target analytes increased by 77% using Sampling Method B from the test portions taken using Sampling Method A. Conclusion Utilizing a sample preparation method that includes a comminution step resulted in an increased number of pathogen detections. Highlights The introduction of a comminution step in the preparation of food samples for detection of three common microbial contaminants resulted in an increase in the rate of detection of natural contaminates in a variety of ready to eat foods. An efficient aseptic process for commutation that can be adapted to a wide range of laboratory settings was identified.

Evaluation of the GENE-UP ®E. coli O157:H7 Method for the Detection of E. coli O157:H7 in Select Foods: Collaborative Study, 2019.03

Background The GENE-UP®E. coli O157:H7 2 (ECO 2) assay (Performance Tested MethodSM 121805) incorporates Fluorescence Resonance Energy Transfer hybridization probes into its proprietary PCR technology for the rapid detection of E. coli O157:H7 in select foods. Objective The purpose of this validation was to evaluate the method’s interlaboratory performance and submit the result to AOAC INTERNATIONAL for adoption as First Action Official MethodSM for the detection of E. coli O157:H7 in select foods. Method The GENE-UP® method was evaluated in a multi-laboratory study as part of the MicroVal validation process using unpaired test portions for one food matrix, raw milk cheese (Comté, 34% fat, 0.8% salt). The candidate method was compared to the ISO 16654:2001 reference method. Fourteen participants from 13 laboratories throughout the European Union participated. Three levels of contamination were evaluated: a non-inoculated control level (0 colony-forming units (CFU)/test portion), a low contamination level (∼5 CFU/test portion), and a high contamination level (∼10 CFU/test portion). Data from that study were analyzed according to the Probability of Detection (POD) statistical model as presented in the AOAC validation guidelines. The difference in laboratory POD (dLPODC) values with 95% confidence interval across collaborators was calculated for each level between the candidate and reference method results, and between the candidate presumptive and confirmed results. Results The dLPODC values with 95% confidence interval were; 0.00 (–0.04, 0.04), 0.27 (0.04, 0.49), and 0.17 (0.01, 0.33) for the non-inoculated, low and high contamination levels respectively. Conclusions The dLPODC results indicate a significant difference between the candidate method and the reference method for both the low and high contamination levels, with the candidate method producing higher recovery of the target organism at both levels. Highlights The GENE-UP E. coli O157:H7 assay provides industry with a rapid, accurate detection method for E. coli O157:H7 in a broad range of foods.