University of Oulu Biomedical Engineering Master’s Thesis Guide: 522987S, 30 ECTS, Laturi, Maturity Test and Biomedical Data Research

Quick answer: what is the University of Oulu Biomedical Engineering thesis route?

The current Biomedical Engineering international Master’s programme at the University of Oulu is a 120 ECTS, two-year Master of Science (Technology) programme in the Faculty of Information Technology and Electrical Engineering (ITEE). In the 2026–2027 Study Guide, the exact thesis is 522987S Master’s Thesis in Biomedical Engineering, 30 ECTS. The current course page places it in Advanced Studies, uses the assessment scale 1–5/FAIL, and shows the 2026–2027 implementation 522987S-3002. The thesis is completed independently under supervision, assessed by the supervisor and a second examiner/reviewer, and the Degree Programme Committee approves the assessment. Students also complete a maturity test. The final thesis route runs through Laturi, with the degree application completed in Peppi.

There is one important time boundary. The University of Oulu is moving to a new 2027–2030 three-year curriculum cycle from 1 August 2027. This guide therefore explains the current 2026–2027 regime verified in September 2026. A student entering in autumn 2027 should recheck the new Study Guide before treating any course code, implementation detail or process date as final.

1. Make sure you are using the correct Biomedical Engineering degree

The target here is the ITEE Master of Science (Technology) Biomedical Engineering programme. The programme combines information technology, electrical engineering and medical-technology expertise, with Faculty of Medicine collaboration, but the degree object remains the ITEE MSc (Technology) programme. This distinction matters because the University also publishes a Faculty of Medicine/MHSc Biomedical Engineering thesis route with a different course code. Do not mix the two because their administrative instructions are not interchangeable.

2. The current thesis is 522987S and it is 30 ECTS

The current 2026–2027 Peppi page identifies 522987S Master’s Thesis in Biomedical Engineering as a 30 ECTS Advanced Studies course. The rendered current page shows implementation 522987S-3002, scheduled from 1 August 2026 to 31 July 2027. It places the work in the second year of MSc studies and states that compulsory advanced studies precede the thesis. Finnish and English are listed as teaching languages.

This is the object that should control planning for the present curriculum. The similarly named 580213S course belongs to the Faculty of Medicine/MHSc route and is not a substitute for 522987S.

3. Understand how the 120 ECTS programme positions the thesis

The current ITEE programme description states that the full Master’s degree is 120 ECTS and contains compulsory advanced studies, the 30 ECTS Master’s thesis, language studies and optional studies according to the personal study plan. The thesis is therefore a major part of the degree but not the whole final-year workload. Use the current PSP and Peppi structure to check what remains alongside the thesis rather than assuming that every course near the thesis page is part of the thesis package.

4. The thesis is independent work, but it is not unsupervised work

The course description says the student defines the thesis content under supervisor guidance and carries out the work independently. The BME thesis instructions emphasize that the purpose is to develop independent and critical engineering/research competence. In practice, the supervisor is not expected to perform the analysis for the student, but supervision should still be used to verify the question, scope, data access, method, technical design and readiness for final submission.

A useful working rule is to arrive at each supervision meeting with a concrete decision to review: the research question, measurement protocol, data split, model baseline, experiment design, device architecture or interpretation of a result. This keeps supervision focused on scientific decisions rather than only on document formatting.

5. Start the process in Laturi with a short research plan

The programme-specific BME instructions direct students to start the thesis process in Laturi. They require a short research plan of maximum one A4 page. The plan should identify the purpose of the work, the methods/materials/devices/software that will be used, and the expected results. The supervisor or supervisors approve the plan in Laturi.

One page is short, so the plan has to make the research logic visible. A good version should state what is being tested or developed, what evidence will answer the question, what the main comparison or evaluation criterion is, and what resources or data are essential. If a dependency such as hospital data, a sensor, company firmware or annotation access is not yet secured, write it as a dependency rather than pretending it is available.

6. A technical assignment still needs an academically evaluable question

Biomedical Engineering theses often involve building or improving something: software, a sensing system, an analysis pipeline, an imaging method, a machine-learning model or a measurement device. A build task alone is not yet an adequate research claim. Define what the solution is expected to improve, how that improvement will be measured, and against what baseline or reference.

For example, “develop a wearable monitoring system” can become a thesis question about measurement accuracy under defined movement conditions, power-performance trade-offs, synchronization reliability, signal quality or prediction performance. The thesis then becomes assessable because the engineering artifact is connected to evidence.

7. The programme does not prescribe one research method

The current BME curriculum covers biomedical sensors and instrumentation, electronics, software, wireless communications, digital signal and image processing, machine learning, cloud services, medical-device regulation, anatomy and physiology, biomechanics, biomaterials and medical imaging. It also contains laboratory work, data analysis, algorithm design and project work. This means a BME thesis may be experimental, computational, design-oriented or a combination.

Do not force every topic into a survey, interview or machine-learning template. The method should follow the actual research problem. A sensor thesis may need calibration and repeatability tests; an imaging thesis may need segmentation or reconstruction evaluation; an ML thesis may need careful validation and leakage control; a biomaterials thesis may need laboratory procedures and statistical comparison.

8. Define the unit of evidence before collecting or splitting data

In health technology, the spreadsheet row is often not the independent unit. One participant can generate thousands of signal windows, repeated images or several visits. If the intended claim is performance on unseen people, putting windows from the same person into both training and test sets can make performance look better than real deployment performance.

Write down the unit at which the conclusion is supposed to generalise: participant, device, session, hospital, sample, image, time period or another level. Then design the split and statistical analysis around that unit. This decision should be made before extensive model tuning.

9. For machine learning, separate development from final evaluation

If the thesis uses machine learning, preprocessing, feature selection, hyperparameters, thresholds and model architecture should be chosen using development data only. A final hold-out set loses its meaning if it is repeatedly checked while decisions are still being made. Cross-validation can be useful, but the resampling structure has to preserve subject or site independence when the research claim requires it.

Report more than one convenient metric when the task demands it. Accuracy can be misleading under class imbalance. Depending on the question, sensitivity, specificity, precision, recall, predictive values, ROC/PR measures, calibration, error distributions and confidence intervals may be more informative. Explain the denominator behind every headline result.

10. For biomedical signals, acquisition is part of the scientific method

A thesis using ECG, EEG, PPG, EMG or other physiological measurements should describe how the signal was obtained, not only how it was analysed. Record the sensor or device, placement, sampling settings, synchronization, recording conditions, firmware or acquisition software and any calibration or quality-control procedure that could affect the output.

Preprocessing is also part of the method. Filtering, resampling, artefact rejection, segmentation, normalization and missing-data handling can change results. Keep the parameters and order of operations reproducible. If a preprocessing choice is changed after looking at model performance, document that development history instead of presenting it as if it had been fixed in advance.

11. For imaging work, preserve patient and acquisition provenance

Medical imaging can generate many slices, frames or derived images from one participant. Image-level random splitting can therefore create the same type of dependency problem as signal-window splitting. Keep identifiers that allow patient- or study-level grouping during analysis while storing direct identifiers separately according to the approved data-governance arrangement.

Also record scanner or modality, protocol, reconstruction settings and relevant preprocessing. A model may learn site- or scanner-specific characteristics rather than the target biological pattern. If data come from more than one device or site, consider whether a held-out site/device analysis is possible and state clearly what external generalisation remains untested.

12. A prototype is not automatically a validated medical device

The programme includes medical-device regulation, but a Master’s thesis prototype should be described at the level actually supported by the evidence. A sensor that functions in laboratory conditions demonstrates one kind of feasibility. A model that performs well retrospectively demonstrates another. Neither result automatically proves clinical safety, diagnostic effectiveness, usability or regulatory conformity.

In the Discussion, separate technical feasibility, analytical performance, clinical performance and clinical utility. If the thesis did not test one of those layers, say so. Careful limitation language is stronger academically than an unsupported claim.

13. Use the BME research plan to expose technical risks early

Because the required plan asks for methods, materials, devices and software, use it to identify dependencies that could stop the thesis. Examples include unavailable equipment, proprietary APIs, licensing restrictions, insufficient sample size, missing reference labels, delayed ethics review, lack of permission to use company data or a dataset that cannot legally be moved to the intended computing environment.

Add a fallback path for the highest-risk dependency. A thesis is easier to finish when the research question can still be answered with a reduced but defensible scope if a partner dataset or device arrives late.

14. The examiner chain is programme-specific

The current thesis course states that the Degree Programme Committee approves the supervisor, reviewers and thesis content. The programme-specific BME page states that the supervisor and second examiner evaluate the thesis in Laturi. They first discuss and agree on the evaluation; the supervisor enters the evaluation and the second examiner joins it. The Degree Programme Committee then accepts the grade.

Do not replace this chain with examiner rules from Computer Science and Engineering or another faculty. The University publishes some nearby programme-specific eligibility rules, but they are not evidence that the same rank requirements apply to BME. For BME, follow the BME supervisor list and current programme instructions.

15. The grading scale is 1–5/FAIL, with a published BME score mapping

The current Peppi course uses 1–5/FAIL. The BME thesis page publishes an evaluation scale in which total scores 10–15 map to grade 1, 16–21 to 2, 22–28 to 3, 29–35 to 4, and 36–42 to 5. Use the current evaluation instructions when planning the quality level of the work rather than waiting for the final week to discover the assessment criteria.

The rubric does not turn a technically complex thesis into a high grade automatically. The manuscript still has to make the problem, method, evidence, reasoning and limitations understandable to examiners who did not build the system.

16. Do not invent a separate BME thesis seminar or seminar ECTS

The current public BME-specific 2026–2027 programme, thesis-course and thesis-process sources do not establish a separate credited BME thesis-seminar course for this MSc Technology route. Some other Oulu programmes have explicit seminar requirements, but those rules cannot be transferred to BME.

This does not prove that no presentation, group meeting or internal seminar activity can ever be used by supervisors. It only means the public evidence does not support publishing a separate seminar code or extra ECTS as a BME degree requirement. If your current PSP, Moodle workspace or updated Peppi structure shows a seminar, follow that current record.

17. The maturity test is required and BME uses the E-exam route

The BME-specific instructions say students must complete a maturity test, and the supervisor creates it as an E-exam for completion in Examinarium. University-wide rules say the maturity test carries 0 ECTS and is assessed pass/fail. It demonstrates familiarity with the thesis subject, while language assessment depends on the student’s previous education and whether Finnish/Swedish proficiency was already demonstrated at the lower-degree stage.

In an English-taught programme, the maturity test is generally written in English unless the Finnish/Swedish language-of-schooling rules create a different requirement. Do not guess a BME-specific maturity course code from a neighbouring programme; register using the current Peppi implementation and supervisor instructions.

18. Plan the submission date backwards from the Degree Programme Committee

The University publishes a current 2026–2027 ITEE Biomedical Engineering graduation timetable. It separates the Laturi thesis deadline, evaluation deadline, thesis/TOTO committee meeting, degree-certificate application deadline and graduation date. This means “my thesis is finished” and “I can graduate this month” are not the same event.

Before promising a graduation date, check the current timetable and reserve time for supervisor review, corrections, formal evaluation, committee processing and the maturity test. The published table is also a reminder that these dates can change by academic year, so use the live page rather than an old screenshot.

19. Laturi is the formal thesis process, not just a file-upload box

Laturi is the University of Oulu system for launching, supervising, monitoring, evaluating, approving and publishing theses. The BME process uses it for the research plan and final thesis workflow. The final document is submitted in PDF/A for assessment and archiving after the supervisor considers it ready.

Treat the Laturi submission as a controlled release. Freeze the final manuscript, verify title and metadata, ensure figures/tables match the final analysis, and remove content that should not appear in a public thesis. Keep the submitted PDF distinct from working versions so that later edits do not create uncertainty about what the examiners actually evaluated.

20. Turnitin checks text similarity; it does not validate the research method

For University of Oulu theses, plagiarism/similarity checking is part of the Laturi process. The University also provides a 2026 Moodle Turnitin environment for students who want to check citations and textual similarity before final Laturi submission.

A low similarity score does not prove that a dataset split is independent, a sensor is calibrated or a statistical conclusion is valid. Originality control and scientific validity are different quality gates. Review both.

21. Company theses must be designed for a public final document

Laturi guidance states that Master’s theses must not contain secret material. Trade or professional secrets can belong only in background material, and their handling should be agreed in advance with the supervisor and, for commissioned work, the commissioning party. This is especially important in BME because company projects may involve proprietary firmware, device architecture, algorithms or commercial datasets.

Resolve publication boundaries when the topic is agreed. Decide what can be described publicly, what data can be shown in aggregate, what diagrams need redaction or replacement, and whether the research question can still be defended without revealing protected information.

22. Personal-data planning must happen before processing starts

The University’s current data-protection policy is explicit for theses. If personal data will be processed, the student must record the lifecycle of that processing in the research plan before processing starts. Data protection and information security must be addressed, a risk assessment and possible data-protection impact assessment must be considered, and research subjects must receive the required privacy notice.

Use data minimisation. Collect only what the academic objective needs. Anonymise or pseudonymise where possible, but remember that pseudonymised data can still be personal data. Keep the re-identification key, if one exists, separate from routine analysis files and restrict access according to the approved arrangement.

23. Ethics depends on the actual project, not the programme label

A BME thesis can range from software testing with non-personal public data to participant measurements or medical research. The ethics route therefore changes with the design. The University’s Human Sciences Ethics Committee reviews non-medical human-sciences research on request under TENK principles, while statutory medical-research ethical advance evaluation belongs to the regional medical-research ethics route.

Do not write “ethics approval was not needed because this is an engineering thesis.” Equally, do not claim that every BME thesis requires ethics-committee approval. Classify the project early with the supervisor and relevant institutional guidance, and obtain any required review before the activity that requires it begins.

24. Data management is part of research quality

University responsible-research guidance requires planning for how research data are collected, stored, shared and preserved, and how reproducibility will be supported. Authorship, ownership, access rights, conditions of use and long-term preservation should be agreed early. These questions matter for both university and company data.

For practical thesis work, maintain a data manifest that links analysis IDs to source files, session/device information, processing status and label provenance without unnecessarily copying direct identifiers. Keep code, configuration and data-version information sufficient to regenerate the results that appear in the thesis.

25. Record software, firmware and model provenance

A biomedical system can change without the dataset filename changing. Record the relevant software version, package environment, firmware, device revision, model configuration and important random seeds. If a model checkpoint generated the reported table, preserve an identifier or checksum for that checkpoint. If a figure comes from a specific script, keep that script with the final analysis record.

This level of provenance prevents a common end-stage problem: the thesis contains a result that cannot be reproduced because the code, model or preprocessing changed after the figure was exported.

26. Keep conclusions at the level supported by evidence

A retrospective model evaluation can support a statement about performance in the evaluated dataset and validation design. It does not automatically support a diagnostic claim. A laboratory prototype can support feasibility under tested conditions; it does not automatically support clinical deployment. A correlation can show association; it does not establish causality or agreement.

Before finalising the Discussion, take every strong conclusion and point to the exact experiment, dataset and metric that supports it. If the connection requires an assumption, either justify the assumption or weaken the wording.

27. Write for the second examiner, not only for your supervisor

Your supervisor may already know the hardware, codebase, dataset and project history. The second examiner may not. The thesis must therefore explain enough context for an informed reader to reconstruct why the problem matters, what was done, how the evidence was generated and why the conclusion follows.

Define abbreviations, describe preprocessing and exclusions, give sample counts at the correct level, identify baselines and report limitations. Do not leave essential reasoning only in code comments, a company meeting, laboratory notebook or oral discussion.

28. Use a reproducibility spot check before Laturi submission

Select at least one headline figure, table or metric and reproduce it from the frozen source data and final analysis configuration. Verify participant/sample counts, exclusion logic, preprocessing, split membership, model or device version and the final numeric value. If the result needs undocumented manual changes, the analysis is not ready to be archived as the evidence behind the thesis.

For hardware work, the equivalent check can be a documented rerun of a core measurement under the stated configuration. For imaging or signal work, it can be regeneration of a representative result from raw or approved intermediate data.

29. A practical 12-step University of Oulu BME workflow

1. Confirm that your degree is the ITEE MSc (Technology) Biomedical Engineering programme and check the current PSP. 2. Confirm that 522987S, 30 ECTS is still your current thesis object. 3. Agree a topic and suitable supervisor(s). 4. Start Laturi and submit the short research plan. 5. Resolve equipment, data access, company-publication boundaries, privacy and ethics requirements before dependent work begins. 6. Define the research question, evidence unit and evaluation plan. 7. Collect/build/analyse with versioned data, software, firmware and configurations. 8. Use supervisor checkpoints to test scope and evidence, not only writing style. 9. Freeze the final analysis and write conclusions within the validated domain. 10. Complete the BME maturity E-exam and final thesis review. 11. Submit the PDF/A through Laturi in time for the relevant evaluation and Degree Programme Committee cycle. 12. Apply for the degree through Peppi after the required studies and records are complete.

30. Final pre-submission checklist

Confirm the current 522987S / 30 ECTS thesis object, supervisor and second examiner route, supervisor-approved research plan, current BME evaluation criteria, maturity-test arrangement, and the live ITEE graduation timetable. Verify data/ethics permissions, privacy notice where needed, public-safe company content, reproducible methods, correct independent unit for analysis, final code/device/model version, limitations, PDF/A formatting, Laturi metadata and Turnitin workflow.

Then verify the temporal boundary one more time. If your thesis work or enrolment moves into the 2027–2030 curriculum after 1 August 2027, recheck Peppi and the BME programme instructions before relying on any code, implementation, deadline or administrative step in this current-regime guide.

Final note on the 2027–2030 curriculum change

The University of Oulu has announced that its first three-year curriculum cycle will cover 2027–2030 and take effect on 1 August 2027. The University also states that transition arrangements will apply. This guide deliberately does not guess what the final BME thesis object will be in that cycle. Until the new programme structure is published and verified, 522987S is documented here as the current 2026–2027 thesis, not as a promise about the future curriculum.