DICOM Anonymization for Clinical Trials: מדריך יישום C# מקיף

using Aspose.Medical.Dicom;
using Aspose.Medical.Dicom.Anonymization;
using System.Collections.Concurrent;
using System.Security.Cryptography;
using System.Text;

public class ClinicalTrialAnonymizer
{
    private readonly string _trialId;
    private readonly ConcurrentDictionary<string, string> _subjectMapping;
    private readonly string _mappingFilePath;
    private readonly string _auditLogPath;

    public ClinicalTrialAnonymizer(string trialId, string dataDirectory)
    {
        _trialId = trialId;
        _mappingFilePath = Path.Combine(dataDirectory, $"{trialId}_subject_mapping.json");
        _auditLogPath = Path.Combine(dataDirectory, $"{trialId}_audit_log.csv");
        _subjectMapping = LoadOrCreateMapping();
        
        InitializeAuditLog();
    }

    private ConcurrentDictionary<string, string> LoadOrCreateMapping()
    {
        if (File.Exists(_mappingFilePath))
        {
            var json = File.ReadAllText(_mappingFilePath);
            var dict = JsonSerializer.Deserialize<Dictionary<string, string>>(json);
            return new ConcurrentDictionary<string, string>(dict);
        }
        return new ConcurrentDictionary<string, string>();
    }

    private void InitializeAuditLog()
    {
        if (!File.Exists(_auditLogPath))
        {
            File.WriteAllText(_auditLogPath, 
                "Timestamp,OriginalFile,AnonymizedFile,SubjectID,Operator,Action\n");
        }
    }

    public string GetOrCreateSubjectId(string originalPatientId)
    {
        return _subjectMapping.GetOrAdd(originalPatientId, _ =>
        {
            int subjectNumber = _subjectMapping.Count + 1;
            return $"{_trialId}-{subjectNumber:D4}";
        });
    }

    public void SaveMapping()
    {
        var json = JsonSerializer.Serialize(
            _subjectMapping.ToDictionary(k => k.Key, v => v.Value),
            new JsonSerializerOptions { WriteIndented = true });
        File.WriteAllText(_mappingFilePath, json);
    }
}

public class TrialAnonymizationResult
{
    public string OriginalPatientId { get; set; }
    public string SubjectId { get; set; }
    public string OriginalFilePath { get; set; }
    public string AnonymizedFilePath { get; set; }
    public DateTime ProcessedAt { get; set; }
    public bool Success { get; set; }
    public string ErrorMessage { get; set; }
}

public TrialAnonymizationResult AnonymizeForTrial(
    string inputPath, 
    string outputDirectory,
    string operatorName)
{
    var result = new TrialAnonymizationResult
    {
        OriginalFilePath = inputPath,
        ProcessedAt = DateTime.UtcNow
    };

    try
    {
        // Load DICOM file
        DicomFile dicomFile = DicomFile.Open(inputPath);
        
        // Get original patient ID and map to subject ID
        string originalPatientId = dicomFile.Dataset.GetString(DicomTag.PatientID) ?? "UNKNOWN";
        string subjectId = GetOrCreateSubjectId(originalPatientId);
        
        result.OriginalPatientId = originalPatientId;
        result.SubjectId = subjectId;

        // Create anonymizer with clinical trial profile
        var profile = CreateClinicalTrialProfile(subjectId);
        var anonymizer = new Anonymizer(profile);
        
        // Anonymize the dataset
        anonymizer.Anonymize(dicomFile.Dataset);
        
        // Generate output filename with subject ID
        string studyDate = dicomFile.Dataset.GetString(DicomTag.StudyDate) ?? "00000000";
        string modality = dicomFile.Dataset.GetString(DicomTag.Modality) ?? "OT";
        string outputFileName = $"{subjectId}_{studyDate}_{modality}_{Guid.NewGuid():N}.dcm";
        string outputPath = Path.Combine(outputDirectory, outputFileName);
        
        // Save anonymized file
        dicomFile.Save(outputPath);
        
        result.AnonymizedFilePath = outputPath;
        result.Success = true;
        
        // Log to audit trail
        LogAuditEntry(inputPath, outputPath, subjectId, operatorName, "ANONYMIZED");
        
        // Save updated mapping
        SaveMapping();
    }
    catch (Exception ex)
    {
        result.Success = false;
        result.ErrorMessage = ex.Message;
        LogAuditEntry(inputPath, "", "", operatorName, $"FAILED: {ex.Message}");
    }

    return result;
}

private void LogAuditEntry(
    string originalFile, 
    string anonymizedFile, 
    string subjectId, 
    string operatorName, 
    string action)
{
    var entry = $"{DateTime.UtcNow:O},{originalFile},{anonymizedFile},{subjectId},{operatorName},{action}\n";
    File.AppendAllText(_auditLogPath, entry);
}

private ConfidentialityProfile CreateClinicalTrialProfile(string subjectId)
{
    // Start with the basic profile for general de-identification
    var options = ConfidentialityProfileOptions.BasicProfile |
                  ConfidentialityProfileOptions.RetainLongitudinalTemporalInformationWithModifiedDates |
                  ConfidentialityProfileOptions.RetainDeviceIdentity;
    
    var profile = ConfidentialityProfile.CreateDefault(options);
    
    // Override specific tags for clinical trial requirements
    // Patient ID becomes the trial subject ID
    profile.SetTagAction(DicomTag.PatientID, 
        new ReplaceAction(subjectId));
    
    // Patient Name becomes anonymized but consistent
    profile.SetTagAction(DicomTag.PatientName, 
        new ReplaceAction($"Subject^{subjectId}"));
    
    // Retain study-level UIDs for longitudinal tracking (but anonymize)
    profile.SetTagAction(DicomTag.StudyInstanceUID, 
        TagAction.ReplaceWithUID);
    
    // Keep clinical trial protocol information
    profile.SetTagAction(DicomTag.ClinicalTrialSponsorName, 
        TagAction.Keep);
    profile.SetTagAction(DicomTag.ClinicalTrialProtocolID, 
        TagAction.Keep);
    profile.SetTagAction(DicomTag.ClinicalTrialProtocolName, 
        TagAction.Keep);
    profile.SetTagAction(DicomTag.ClinicalTrialSiteID, 
        TagAction.Keep);
    profile.SetTagAction(DicomTag.ClinicalTrialSubjectID, 
        new ReplaceAction(subjectId));
    
    return profile;
}

public class MultiSiteTrialAnonymizer
{
    private readonly string _trialId;
    private readonly string _siteId;
    private readonly ClinicalTrialAnonymizer _anonymizer;

    public MultiSiteTrialAnonymizer(string trialId, string siteId, string dataDirectory)
    {
        _trialId = trialId;
        _siteId = siteId;
        
        // Each site has its own mapping file
        string siteDataDir = Path.Combine(dataDirectory, siteId);
        Directory.CreateDirectory(siteDataDir);
        
        _anonymizer = new ClinicalTrialAnonymizer($"{trialId}-{siteId}", siteDataDir);
    }

    public async Task<List<TrialAnonymizationResult>> ProcessSiteSubmission(
        string inputDirectory,
        string outputDirectory,
        string operatorName)
    {
        var results = new List<TrialAnonymizationResult>();
        
        // Create site-specific output directory
        string siteOutputDir = Path.Combine(outputDirectory, _siteId);
        Directory.CreateDirectory(siteOutputDir);
        
        var dicomFiles = Directory.GetFiles(inputDirectory, "*.dcm", SearchOption.AllDirectories);
        
        foreach (var filePath in dicomFiles)
        {
            var result = _anonymizer.AnonymizeForTrial(filePath, siteOutputDir, operatorName);
            results.Add(result);
            
            // Log progress
            Console.WriteLine($"[{_siteId}] Processed: {Path.GetFileName(filePath)} -> {result.SubjectId}");
        }
        
        // Generate site submission manifest
        GenerateSubmissionManifest(results, siteOutputDir);
        
        return results;
    }

    private void GenerateSubmissionManifest(
        List<TrialAnonymizationResult> results, 
        string outputDirectory)
    {
        var manifest = new
        {
            TrialId = _trialId,
            SiteId = _siteId,
            SubmissionDate = DateTime.UtcNow,
            TotalFiles = results.Count,
            SuccessfulFiles = results.Count(r => r.Success),
            FailedFiles = results.Count(r => !r.Success),
            Subjects = results
                .Where(r => r.Success)
                .GroupBy(r => r.SubjectId)
                .Select(g => new
                {
                    SubjectId = g.Key,
                    FileCount = g.Count()
                })
                .ToList()
        };

        string manifestPath = Path.Combine(outputDirectory, "submission_manifest.json");
        string json = JsonSerializer.Serialize(manifest, new JsonSerializerOptions { WriteIndented = true });
        File.WriteAllText(manifestPath, json);
    }
}

public class LongitudinalTrialAnonymizer
{
    private readonly ClinicalTrialAnonymizer _baseAnonymizer;
    private readonly Dictionary<string, List<StudyInfo>> _subjectStudies;

    public class StudyInfo
    {
        public string OriginalStudyUID { get; set; }
        public string AnonymizedStudyUID { get; set; }
        public DateTime OriginalStudyDate { get; set; }
        public DateTime AnonymizedStudyDate { get; set; }
        public int DayOffset { get; set; }
    }

    public LongitudinalTrialAnonymizer(string trialId, string dataDirectory)
    {
        _baseAnonymizer = new ClinicalTrialAnonymizer(trialId, dataDirectory);
        _subjectStudies = new Dictionary<string, List<StudyInfo>>();
    }

    public void AnonymizeWithTemporalConsistency(
        string inputPath,
        string outputDirectory,
        string operatorName)
    {
        DicomFile dicomFile = DicomFile.Open(inputPath);
        
        string patientId = dicomFile.Dataset.GetString(DicomTag.PatientID);
        string subjectId = _baseAnonymizer.GetOrCreateSubjectId(patientId);
        
        string originalStudyUID = dicomFile.Dataset.GetString(DicomTag.StudyInstanceUID);
        DateTime originalStudyDate = ParseDicomDate(
            dicomFile.Dataset.GetString(DicomTag.StudyDate));
        
        // Get or create study info for temporal consistency
        var studyInfo = GetOrCreateStudyInfo(subjectId, originalStudyUID, originalStudyDate);
        
        // Create profile with consistent date shifting
        var profile = CreateLongitudinalProfile(subjectId, studyInfo);
        var anonymizer = new Anonymizer(profile);
        
        anonymizer.Anonymize(dicomFile.Dataset);
        
        // Apply consistent study UID
        dicomFile.Dataset.AddOrUpdate(DicomTag.StudyInstanceUID, studyInfo.AnonymizedStudyUID);
        
        // Apply shifted date
        dicomFile.Dataset.AddOrUpdate(DicomTag.StudyDate, 
            studyInfo.AnonymizedStudyDate.ToString("yyyyMMdd"));
        
        string outputPath = GenerateOutputPath(outputDirectory, subjectId, studyInfo);
        dicomFile.Save(outputPath);
    }

    private StudyInfo GetOrCreateStudyInfo(
        string subjectId, 
        string originalStudyUID, 
        DateTime originalStudyDate)
    {
        if (!_subjectStudies.ContainsKey(subjectId))
        {
            _subjectStudies[subjectId] = new List<StudyInfo>();
        }

        var existingStudy = _subjectStudies[subjectId]
            .FirstOrDefault(s => s.OriginalStudyUID == originalStudyUID);
        
        if (existingStudy != null)
        {
            return existingStudy;
        }

        // Calculate day offset from first study
        int dayOffset = 0;
        if (_subjectStudies[subjectId].Any())
        {
            var firstStudy = _subjectStudies[subjectId].First();
            dayOffset = (originalStudyDate - firstStudy.OriginalStudyDate).Days;
        }

        var newStudy = new StudyInfo
        {
            OriginalStudyUID = originalStudyUID,
            AnonymizedStudyUID = GenerateConsistentUID(originalStudyUID),
            OriginalStudyDate = originalStudyDate,
            AnonymizedStudyDate = new DateTime(2000, 1, 1).AddDays(dayOffset),
            DayOffset = dayOffset
        };

        _subjectStudies[subjectId].Add(newStudy);
        return newStudy;
    }

    private string GenerateConsistentUID(string originalUID)
    {
        // Generate deterministic UID based on original
        using (var sha = SHA256.Create())
        {
            byte[] hash = sha.ComputeHash(Encoding.UTF8.GetBytes(originalUID));
            string hashString = BitConverter.ToString(hash).Replace("-", "").Substring(0, 20);
            return $"2.25.{hashString}";
        }
    }

    private DateTime ParseDicomDate(string dicomDate)
    {
        if (DateTime.TryParseExact(dicomDate, "yyyyMMdd", null, 
            System.Globalization.DateTimeStyles.None, out var date))
        {
            return date;
        }
        return DateTime.MinValue;
    }
}

public class TrialSubmissionReportGenerator
{
    public void GenerateReport(
        string trialId,
        List<TrialAnonymizationResult> results,
        string outputPath)
    {
        var report = new StringBuilder();
        
        report.AppendLine("CLINICAL TRIAL IMAGING DATA ANONYMIZATION REPORT");
        report.AppendLine("================================================");
        report.AppendLine();
        report.AppendLine($"Trial ID: {trialId}");
        report.AppendLine($"Report Generated: {DateTime.UtcNow:yyyy-MM-dd HH:mm:ss} UTC");
        report.AppendLine($"Total Files Processed: {results.Count}");
        report.AppendLine($"Successful: {results.Count(r => r.Success)}");
        report.AppendLine($"Failed: {results.Count(r => !r.Success)}");
        report.AppendLine();
        
        report.AppendLine("SUBJECT SUMMARY");
        report.AppendLine("---------------");
        
        var subjectGroups = results
            .Where(r => r.Success)
            .GroupBy(r => r.SubjectId)
            .OrderBy(g => g.Key);
        
        foreach (var group in subjectGroups)
        {
            report.AppendLine($"  {group.Key}: {group.Count()} files");
        }
        
        report.AppendLine();
        report.AppendLine("ANONYMIZATION PROFILE");
        report.AppendLine("---------------------");
        report.AppendLine("  Base Profile: DICOM PS 3.15 Basic Application Level Confidentiality Profile");
        report.AppendLine("  Modifications: Retain Longitudinal Temporal Information with Modified Dates");
        report.AppendLine("  Subject ID Format: [TrialID]-[SequentialNumber]");
        report.AppendLine();
        
        if (results.Any(r => !r.Success))
        {
            report.AppendLine("PROCESSING ERRORS");
            report.AppendLine("-----------------");
            foreach (var failed in results.Where(r => !r.Success))
            {
                report.AppendLine($"  File: {failed.OriginalFilePath}");
                report.AppendLine($"  Error: {failed.ErrorMessage}");
                report.AppendLine();
            }
        }
        
        report.AppendLine("CERTIFICATION");
        report.AppendLine("-------------");
        report.AppendLine("This report certifies that all DICOM files listed above have been");
        report.AppendLine("processed through an automated anonymization pipeline in compliance");
        report.AppendLine("with HIPAA Safe Harbor de-identification requirements.");
        
        File.WriteAllText(outputPath, report.ToString());
    }
}

public class Program
{
    public static async Task Main(string[] args)
    {
        // Initialize metered license
        Metered metered = new Metered();
        metered.SetMeteredKey("your-public-key", "your-private-key");

        string trialId = "ONCO-2025-001";
        string siteId = "SITE-NYC";
        string dataDir = @"C:\ClinicalTrials\Data";
        string inputDir = @"C:\ClinicalTrials\Incoming\Site_NYC";
        string outputDir = @"C:\ClinicalTrials\Anonymized";
        string operatorName = "DataManager_JSmith";

        // Process site submission
        var siteAnonymizer = new MultiSiteTrialAnonymizer(trialId, siteId, dataDir);
        var results = await siteAnonymizer.ProcessSiteSubmission(inputDir, outputDir, operatorName);

        // Generate submission report
        var reportGenerator = new TrialSubmissionReportGenerator();
        reportGenerator.GenerateReport(
            trialId,
            results,
            Path.Combine(outputDir, siteId, "anonymization_report.txt"));

        Console.WriteLine($"Processed {results.Count} files");
        Console.WriteLine($"Success: {results.Count(r => r.Success)}");
        Console.WriteLine($"Failed: {results.Count(r => !r.Success)}");
    }
}