| 1 Definition | 2 Functional Requirements | 3 Syntax | 4 Semantics |
1 Definition
Types of Data Processing that an AIH Data Processing AIM can be performed on AIH Data..
2 Functional Requirements
The Functional Requirements are organised into Common Definitions and sections specific to ECG, EEG, Medical Image, Genomics domains.
2.1 Common Definitions – Functional Requirements
- The Common Definitions shall define the permissible top-level Domain values (ECG, EEG, Medical Image, Genomics).
- The Common Definitions shall enable routing of validation and processing logic based on Domain.
- The Common Definitions shall allow Feature Class to specify the category of features (e.g., Morphological, Temporal, Frequency, Statistical, HRV, Spectral bands, ERP components).
- The Common Definitions shall validate Features as a non-empty array of strings with unique items.
- The Common Definitions shall support Algorithm as either a string identifier or an object with Name, Version, and Params.
- The Common Definitions shall support Algorithms as an array whose items are either a string identifier or an Algorithm object.
- The Common Definitions shall capture provenance using Trace.
2.2 ECG Processing Type – Functional Requirements
- The ECG Processing Type shall support operations: Pre-processing, Beat Detection, Wave Delimitation, Feature Extraction, Arrhythmia Classification, Heart Rate Variability (HRV) Analysis.
- The ECG Processing Type shall validate Operation against ECG-specific enumerations (e.g., RPeak Detection, QRS Detection, QT Measurement, ST Deviation, HRV Time Domain, HRV Frequency Domain, Beat Classification, Baseline Wander Removal, Denoising, Arrhythmia Detection).
- The ECG Processing Type shall validate Method against ECG techniques (e.g., Pan Tompkins, Wavelet Transform, Savitzky Golay, Adaptive Thresholding, Template Matching, Lomb Scargle, Welch PSD, Median Filter, CNN, RNN, Transformer).
- The ECG Processing Type shall allow Targets to include waveform components P, QRS, T, ST.
2.3 EEG Processing Type – Functional Requirements
- The EEG Processing Type shall support operations: Pre-processing, Epoching, Feature Extraction, Source Localisation, Classification.
- The EEG Processing Type shall validate Operation against EEG-specific enumerations (e.g., Band Power, Spectral Analysis, Event Related Potential, Artifact Removal, Independent Component Analysis (ICA) Decomposition, Source Localisation, Sleep Stage Classification, Epoching, Connectivity Analysis, Time Frequency Analysis).
- The EEG Processing Type shall validate Method against EEG techniques (e.g., FFT, Welch PSD, Morlet Wavelet, ICA, CSP, sLORETA, Beamforming, Notch Filter, Bandpass Filter, DCNN, LSTM).
2.4 MedicalImaging Processing Type – Functional Requirements
- The Medical Imaging Processing Type shall support operations: Pre-processing, Segmentation, Registration, Detection Classification, Reconstruction, Quantification.
- The Medical Imaging Processing Type shall validate Operation against imaging-specific enumerations (e.g., Segmentation, Registration, Denoising, Enhancement, Lesion Detection, Classification, Quantification, Reconstruction, Motion Correction, Feature Extraction).
- The Medical Imaging Processing Type shall validate Method against imaging techniques (e.g., Otsu Thresholding, UNet, ResNet, FLIRT, ANTs, SIFT, SURF, Non Local Means, Bilateral Filter, Histogram Equalisation, CNN, Transformer, LevelSet, GraphCut).
2.5 Genomic Processing Type – Functional Requirements
- The Genomic Processing Type shall support operations: Alignment, Quality Control, Trimming, Variant Calling, Variant Filtration, Annotation, Expression Quantification, Differential Expression, Peak Calling, Assembly, Phasing, Normalization.
- The Genomic Processing Type shall validate Method against genomics techniques (e.g., BWA, Bowtie2, Minimap2, STAR, HISAT2, GATK. Haplotype Caller, Free Bayes, bcftools, Samtools, FastQC, Cutadapt, Trimmomatic, DESeq2, edgeR, Salmon, Kallisto, MACS2, SPAdes, Trinity).
3 Syntax
https://schemas.mpai.community/AIH1/V1.0/data/AIHDataProcessingType.json
4 Semantics
| Label | Description |
| Header | Health Data Processing Types Header |
| – Standard – HealthDataProcessingTypes | The characters AIH-HDP-V |
| – Version | Major version – 1 or 2 characters |
| – Dot-separator | The character . |
| – Subversion | Minor version – 1 or 2 characters |
| Common Definitions | |
| Domains | Top-level processing areas covered by this taxonomy. |
| Algorithm | Identifier (string) or object (Name, Version, Params). Use ID when minimal; object when detailed. |
| FeatureClass | Class of features (e.g., Morphological, Temporal, Frequency, Statistical, HRV, Spectral bands, ERP components). |
| Features | Examples per domain: ECG → QRS_Duration, RR_Interval, SDNN, RMSSD; EEG → PSD bands, ERP amplitudes; Imaging → Radiomics features. |
| Trace | Capture source/time provenance (MPAI Trace). Trace.SourceAIM → AIF/V3.0/data/AIMInstance.json; Trace.Time → OSD/V1.5/data/Time.json. |
| DescrMetadata | Descriptive metadata for human-readable context (title, description, notes). |
| Domains – ECG | Electrocardiogram signal processing domain. |
| –– Preprocessing | Noise/artifact mitigation before analysis. |
| –– BeatDetection | Detection of heart beats and R-peaks. |
| –– WaveDelimitation | Identification of P/QRS/T boundaries and ST segments. |
| –– FeatureExtraction | Derivation of morphological/temporal/frequency features. |
| –– ArrhythmiaClassification | Automated classification of rhythm abnormalities. |
| –– HRVAnalysis | Time/frequency/nonlinear analysis of heart-rate variability. |
| Methods – ECG | Techniques for ECG domain operations. |
| –– Preprocessing | BaselineWanderRemoval; PowerlineNotch; BandpassFiltering; MotionArtifactSuppression; WaveletDenoising. |
| –– BeatDetection | QRSDetection; RPeakDetection. |
| –– WaveDelimitation | PWaveSegmentation; QRSTDelimitation; TWaveEndDetection. |
| –– FeatureExtraction – FeatureClass | Morphological; Temporal; Frequency; Statistical; HRV. |
| –– FeatureExtraction – Features | Examples: QRS_Duration, RR_Interval, PSD bands, SDNN, RMSSD. |
| –– ArrhythmiaClassification | RuleBased; ClassicalML; DeepLearning. |
| –– HRVAnalysis | TimeDomain; FrequencyDomain; Nonlinear. |
| Targets – P | Atrial depolarisation wave. |
| Targets – QRS | Ventricular depolarisation complex. |
| Targets – T | Ventricular repolarisation wave. |
| Targets – ST | Segment between S end and T start. |
| Domains – EEG | Electroencephalography signal processing domain. |
| –– Preprocessing | Filtering, re-referencing, artifact removal. |
| –– Epoching | Segmentation of continuous EEG around events. |
| –– FeatureExtraction | Spectral/ERP/time–frequency/connectivity features. |
| –– SourceLocalization | Estimation of neural generators from scalp signals. |
| –– Classification | Model-based categorisation of signals or states. |
| Methods – EEG | Techniques for EEG domain operations. |
| –– Preprocessing | Filtering_Bandpass; Filtering_Notch; ReReferencing; Resampling; ArtifactRejection_ICA; ArtifactRejection_Automated. |
| –– Epoching | EventLockedEpochs; BaselineCorrection. |
| –– FeatureExtraction | Spectral; ERP; TimeFrequency; Connectivity. |
| –– FeatureExtraction – FeatureClass | Spectral bands; ERP components; wavelet/STFT; coherence/PLI/PLV. |
| –– SourceLocalization | DipoleFitting; DistributedInverseSolution. |
| –– Classification | ClassicalML; DeepLearning. |
| Domains – Genomics | Genomics/transcriptomics processing domain (e.g., WGS/WES/RNA-seq/ChIP-seq). |
| –– Alignment | Mapping reads to a reference genome/transcriptome. |
| –– QualityControl | Read/coverage quality assessment and reporting. |
| –– Trimming | Adapter/low-quality base removal before alignment. |
| –– VariantCalling | Detection of SNPs/indels/structural variants. |
| –– VariantFiltration | Applying quality/annotation-based filters to variants. |
| –– Annotation | Adding biological/functional context to variants. |
| –– ExpressionQuantification | Quantifying gene/transcript expression from RNA-seq. |
| –– DifferentialExpression | Comparing expression between conditions/groups. |
| –– PeakCalling | Identifying enriched regions (e.g., ChIP-seq peaks). |
| –– Assembly | De novo or reference-guided assembly of sequences. |
| –– Phasing | Inferring haplotypes from genotype/reads. |
| –– Normalization | Scaling/normalizing counts/signals for comparability. |
| Methods – Genomics | Techniques for genomics operations. |
| –– Alignment | BWA; Bowtie2; Minimap2; STAR; HISAT2. |
| –– QualityControl | FastQC. |
| –– Trimming | Cutadapt; Trimmomatic. |
| –– VariantCalling | GATK.HaplotypeCaller; FreeBayes; bcftools; Samtools. |
| –– Annotation | Examples include ANNOVAR or SnpEff (if used). |
| –– ExpressionQuantification | Salmon; Kallisto. |
| –– DifferentialExpression | DESeq2; edgeR. |
| –– PeakCalling | MACS2. |
| –– Assembly | SPAdes; Trinity. |
| –– Normalization | TPM/RPKM/FPKM or domain-specific approaches (if applicable). |
| –– FeatureExtraction – FeatureClass | Coverage; Variants; Expression; Peak metrics. |
| –– FeatureExtraction – Features | Examples: read_depth, variant_count, TPM, peak_score. |
| Domains – MedicalImaging | Medical image processing domain. |
| –– Preprocessing | Denoising, contrast/bias correction, normalization. |
| –– Segmentation | Partitioning images into anatomical/lesion regions. |
| –– Registration | Spatial alignment within/across modalities or time. |
| –– DetectionClassification | Finding and labeling abnormalities or tissues. |
| –– Reconstruction | Improving or rebuilding images from raw/undersampled data. |
| –– Quantification | Measurement and radiomics feature computation. |
| Methods – MedicalImaging | Techniques for Medical Imaging domain operations. |
| –– Preprocessing | Denoising; ContrastEnhancement; BiasFieldCorrection; Normalization. |
| –– Segmentation | Thresholding; RegionGrowing; ActiveContour; GraphCut; UNet; AttentionUNet; TransformerBased. |
| –– Registration | Rigid; Affine; Deformable; IntensityBased; LandmarkBased. |
| –– DetectionClassification | FeatureBased; ClassicalML; DeepLearning. |
| –– Reconstruction | IterativeReconstruction; SuperResolution; Denoising. |
| –– Quantification | Radiomics; VolumeMeasurement; ShapeAnalysis. |
