REGULATION OF DIGITAL PATHOLOGY AND ARTIFICIAL INTELLIGENCE INTEGRATION IN JAPAN

1. INTRODUCTION

Digital pathology with artificial intelligence (AI) integration involves converting traditional glass-slide tissue examinations into high-resolution digital images and applying AI algorithms for disease detection, tumour classification, cellular analysis and diagnostic assistance. These technologies can improve diagnostic efficiency, support cancer screening and facilitate collaboration between pathology laboratories.

In Japan, digital pathology and AI-assisted diagnosis are governed through a combination of the Act on Securing Quality, Efficacy and Safety of Products Including Pharmaceuticals and Medical Devices (PMD Act), Medical Practitioners Act, Medical Care Act, Act on Clinical Laboratory Technicians and the Act on the Protection of Personal Information (APPI).

Japan does not have a separate comprehensive statute exclusively regulating AI-based digital pathology. Instead, existing medical-device, professional-responsibility, clinical-laboratory and data-protection requirements apply to different components of the technology.

2. LEGAL AND REGULATORY FRAMEWORK

A. Pharmaceuticals and Medical Devices Act (PMD Act)

The PMD Act is the principal legislation governing AI software intended for medical diagnosis.

Under this framework:

AI software intended to assist pathological diagnosis may qualify as Software as a Medical Device (SaMD).

Software classification depends on its intended medical purpose and associated patient risks.

Relevant Class II or higher products may require third-party certification or regulatory approval.

Manufacturers must demonstrate appropriate quality, safety and performance.

Post-market safety monitoring and regulatory compliance obligations apply.

The Pharmaceuticals and Medical Devices Agency (PMDA) evaluates relevant medical-device applications and provides regulatory guidance.

Importantly, not every digital pathology viewer is automatically a regulated medical device. Classification depends on its actual intended functionality.

B. Medical Practitioners Act

Article 17 prohibits persons who are not physicians from practising medicine.

Consequently, AI systems cannot independently replace the legally responsible physician in making medical diagnoses. AI-generated pathological findings must be evaluated within the applicable professional framework.

C. Act on Clinical Laboratory Technicians

The Act on Clinical Laboratory Technicians (Act No. 76 of 1958) regulates professional activities involving pathological examinations.

It establishes qualifications and professional responsibilities for clinical laboratory technicians. Their work must remain within legally permitted professional functions and the applicable physician-supervision framework.

D. Medical Care Act

The Medical Care Act establishes requirements concerning medical institutions and the appropriate management of clinical laboratory testing.

Article 15-2 requires medical institutions conducting specimen examinations to comply with applicable standards concerning facilities, management structures and examination accuracy.

These requirements are particularly relevant when digital pathology systems are integrated into hospital laboratories.

3. AI INTEGRATION AND VALIDATION REQUIREMENTS

AI-based digital pathology systems must undergo appropriate technical and clinical validation.

Important compliance requirements include:

Diagnostic accuracy: Testing sensitivity, specificity and diagnostic performance against appropriate reference standards.

Dataset quality: Ensuring training and validation datasets adequately represent relevant patient populations and tissue characteristics.

Algorithm reliability: Evaluating false-positive and false-negative results.

Clinical validation: Demonstrating that the software performs reliably within its intended clinical application.

Human oversight: Maintaining appropriate physician review of AI-generated findings.

Software updates: Evaluating significant algorithm changes and determining whether regulatory approval or certification variations are necessary.

Cybersecurity: Protecting digital slides, diagnostic reports and connected medical systems against unauthorised access.

A system validated for one cancer type or specimen category should not automatically be treated as clinically validated for every pathological application.

4. PERSONAL DATA PROTECTION AND DIGITAL SLIDE MANAGEMENT

Digital pathology images frequently contain identifiable patient information.

Under the APPI, medical information requires appropriate protection.

Healthcare institutions should implement:

Lawful handling of patient information.

Appropriate access restrictions.

Secure storage and transmission.

Pseudonymisation or anonymisation where appropriate.

Controls over secondary research use.

Appropriate arrangements for external cloud-service providers.

Measures addressing international data transfers.

The Japanese Society of Pathology also published guidance in November 2022 concerning the handling of Whole Slide Images (WSI), distinguishing clinical, research and educational applications.

These guidelines provide important operational guidance for institutions adopting digital pathology.

5. IMPORTANT JUDICIAL PRECEDENTS

There is no established Japanese Supreme Court judgment specifically determining liability for AI-generated digital pathology errors. Nevertheless, the following judicial decisions provide relevant direct or analogous legal principles.

1. Tokyo District Court Judgment, 12 April 2018

Direct authority concerning pathological diagnostic disputes; analogous to AI-assisted pathology.

The dispute involved fine-needle aspiration cytology in which a cytotechnologist classified a specimen as benign, while the pathologist suspected malignancy. Subsequent examination of the surgically removed tissue did not identify cancer cells. The court dismissed the patient's damages claim.

The case illustrates the importance of professional roles, diagnostic judgment and the evaluation of pathological examination procedures.

2. Fukuoka High Court, Miyazaki Branch, 1 February 2000

Direct authority concerning pathological specimen handling.

The court considered the removal of a liver specimen during an autopsy without appropriate informed consent. The acquisition of the specimen was found unlawful.

The decision demonstrates the importance of consent and lawful specimen management.

3. Tokyo District Court Judgment, 24 November 2000

Direct authority concerning pathological specimens.

The dispute concerned the scope of consent for retaining bone and bone-marrow specimens. The court required the return of specimens, including paraffin-embedded tissue blocks, because appropriate consent had not been obtained.

This decision is relevant to the governance of stored digital pathology material and associated tissue specimens.

4. Tokyo District Court Judgment, 30 August 2001

Direct authority concerning consent and specimen retention.

The court considered whether the consent obtained for retaining organs extended to bone and bone-marrow specimens. Unlike the preceding dispute, the court accepted that the consent covered the retention in question.

The decision demonstrates the importance of clearly defining the scope of specimen-related consent.

5. Tokyo District Court Judgment, 5 February 2007

Analogous authority concerning pathological diagnostic accuracy.

The litigation concerned an allegedly incorrect pathological diagnosis following an anatomical autopsy. The court concluded that the final diagnosis was correct and found no breach of the hospital's duty of care.

This decision is relevant when evaluating whether an alleged AI-supported diagnostic error amounts to legally actionable negligence.

6. Supreme Court Judgment, 11 November 2003

Analogous authority concerning medical negligence and diagnostic responsibility.

The Supreme Court considered a physician's failure to transfer a patient promptly to an appropriate medical institution despite circumstances indicating a potentially serious condition.

The Court recognised liability where negligence deprived the patient of a substantial possibility of avoiding serious consequences.

The reasoning is relevant to situations where healthcare professionals rely excessively on AI outputs and fail to pursue necessary additional investigations.

6. LIABILITY AND ENFORCEMENT MECHANISMS

Liability for digital pathology failures may arise at several levels.

A. Physician liability: A pathologist may face civil liability for failing to exercise appropriate professional judgment or ignoring significant contradictory findings.

B. Hospital liability: Medical institutions may face liability for inadequate supervision, defective laboratory procedures or insufficient safety management.

C. Manufacturer liability: Manufacturers may face regulatory or civil consequences for defective medical devices, misleading claims or failures to satisfy applicable legal obligations.

D. Data-protection liability: Unauthorised disclosure or improper processing of patient images may trigger consequences under the APPI.

E. Institutional enforcement: Regulatory authorities may require corrective action, investigate compliance failures or impose applicable administrative measures.

7. PRACTICAL COMPLIANCE REQUIREMENTS

Japanese hospitals and pathology laboratories should establish:

Documented AI validation protocols.

Qualified pathologist supervision.

Digital-slide quality-control procedures.

Audit trails for AI-generated findings.

Cybersecurity and patient-data protection systems.

Software-version and algorithm-change documentation.

Incident-reporting and corrective-action procedures.

Clear allocation of responsibilities between hospitals, pathologists and technology suppliers.

8. CONCLUSION

Japan regulates AI-integrated digital pathology through an interconnected legal framework involving the PMD Act, Medical Practitioners Act, Medical Care Act, Act on Clinical Laboratory Technicians and APPI.

The principal regulatory objectives are diagnostic reliability, patient safety, professional accountability, lawful handling of pathological specimens and protection of medical information.

Japanese judicial precedents concerning pathological specimen management and diagnostic negligence provide important guidance, although they do not establish AI-specific liability rules.

Ultimately, AI should function as a clinically validated diagnostic-support technology operating under appropriate professional supervision. Hospitals, pathologists and manufacturers must maintain clear accountability, reliable validation procedures and effective patient-data safeguards to ensure legally compliant digital pathology implementation.