University of Turku Biomedical Imaging Master's Thesis Guide: BIMA3240, 40 ECTS, Experimental Research and UTUGradu

Quick answer: what is the Biomedical Imaging thesis route?

The current University of Turku Master’s Degree Programme in Biomedical Imaging is a 120 ECTS, two-year Master of Science programme connected at Turku to the Faculty of Medicine and Institute of Biomedicine and jointly organised with Åbo Akademi University. The controlling current Peppi object is programme 105701, MDP in Biomedical Imaging (MSc), 2025–2027. The thesis is not 30 ECTS. It is BIMA3240 Master’s Thesis in Biomedical Imaging, 40 ECTS, assessed on the 0–5 scale. It is an experimental research project conducted under supervision in a research group. The thesis plan must be accepted by the supervisor, presented in the thesis-plan seminar and approved by the responsible professor before the practical part begins.

1. Use the current standalone Biomedical Imaging programme, not the old track structure

Biomedical Imaging now has its own current Peppi programme object. For thesis planning, use programme 105701 and its 2025–2027 structure rather than older material where Biomedical Imaging appeared as a track under a broader Biomedical Sciences master’s programme. This matters because course packaging, credits and programme rules can change when a track becomes a standalone programme. The current degree is 120 ECTS over two years, and current programme-level and course-level evidence independently agree on the 40 ECTS BIMA3240 thesis route.

2. Understand the 120 ECTS structure before calculating the thesis

Current Peppi places 96–97 ECTS in Major Subject Studies in Biomedical Imaging. This includes 33 ECTS of mandatory major studies, 3–4 ECTS of Professional and Career Development, 20 ECTS of selectable major-subject modules and the 40 ECTS thesis. Students also complete 5 ECTS of language and Finnish-culture studies and enough electives to reach the 120 ECTS degree. Supplementary studies of 0–16 ECTS can be required depending on background. They are not a universal additional block that every student adds beyond 120 ECTS.

3. The thesis is BIMA3240 and it carries 40 ECTS

Peppi identifies one current course unit for the thesis: BIMA3240 Master’s Thesis in Biomedical Imaging, 40 ECTS. This is the assessed thesis object for the current standalone programme. The 40 credits sit inside the major-subject block and therefore inside the 120 ECTS degree. Do not add 40 credits after calculating 120 ECTS, and do not import a 30 ECTS thesis figure from another University of Turku programme. Peppi schedules BIMA3240 entirely in year two, with 16 ECTS in semester three and 24 ECTS in semester four.

4. This is an experimental research thesis

The current programme description says that the thesis is based on an experimental research project corresponding to about five to six months of full-time work. BIMA3240 requires practical laboratory work or data generation, analysis and a written scientific thesis. This makes the programme-specific route materially different from a generic literature-only master’s thesis. Literature review is an important part of the work, but it supports the experimental project. Topic, design and analysis still vary widely across microscopy, medical imaging, image informatics and other biomedical-imaging research environments.

5. Finding the thesis project is part of the student’s responsibility

BIMA3240 states that students are responsible for finding their thesis projects. The work is normally conducted under supervision in a research group, so project discovery should start early enough to identify a suitable group, discuss feasibility and understand what equipment, data, samples, approvals and analytical support actually exist. A project should not be selected only because its title sounds clinically important or technically advanced. It needs to fit the programme, the available supervision and a realistic five-to-six-month experimental workload.

6. Discuss the subject with the responsible professor before starting

The current thesis course says that the subject must first be discussed with the responsible professor. This is an important programme-specific control because it places academic scope before practical execution. At this stage, clarify the research problem, the main imaging or analysis environment, what evidence can realistically be generated and whether the intended project needs specialist ethics, data, hospital, animal or other approvals. A narrow, feasible project with a clear evidence chain is normally stronger than an ambitious project that cannot be validated within the thesis period.

7. The thesis plan is a hard checkpoint before practical work

Before the practical part begins, the student must submit a thesis plan that has been accepted by the supervisor. The student then presents the plan in the thesis-plan seminar, and the responsible professor must approve it before practical work can begin. Treat this as a real gate. Do not start thesis data generation, participant recruitment, regulated animal procedures or project-specific laboratory work simply because a topic has been informally discussed. The plan should make the research aims, practical route, analysis and applicable ethics or data controls clear enough for approval.

8. The thesis seminars are embedded inside BIMA3240

The current BIMA3240 course includes a thesis-plan seminar, interim thesis progress meetings and a final Master’s thesis seminar. These activities are part of the 40 ECTS thesis workload. They are not a separately credited thesis-seminar sequence. This distinction is important because other Turku programmes package seminars differently. For Biomedical Imaging, do not add separate seminar credits unless your personal study plan contains another course for a different purpose. The thesis results are presented orally in the final thesis seminar.

9. BIMA2113 Scientific Seminar Diary is not the thesis seminar

Peppi also lists BIMA2113 Scientific Seminar Diary, 2 ECTS under Professional and Career Development. It is a separate curriculum object. It should not be described as the thesis-plan seminar or the final BIMA3240 thesis seminar, and its credits should not be added as if BIMA3240 required a separate 2 ECTS thesis-seminar course. This is a useful example of why course names containing the word “seminar” cannot be inherited or interpreted by title alone. The controlling course description must be checked.

10. The workload explains why the thesis must be planned as a research project

BIMA3240 specifies 1080 hours in total. It allocates 270 hours to seminars and planning, including 70 hours of seminar work, 100 hours of thesis planning and 100 hours of literature review. It then allocates 810 hours to the thesis itself, including 600 hours of laboratory work and analysis and 210 hours of writing. These figures show that literature, planning, experimentation, analysis, oral presentation and writing are all explicit parts of the current thesis. Leaving analysis or writing until the experimental work is finished creates unnecessary risk.

11. The written thesis has a conventional scientific structure

The current course description identifies a relevant literature review, aims of the study, materials and methods, results and conclusions as core written components. The thesis should therefore make a traceable argument from background and research gap to aims, experimental design, observations, analysis and interpretation. The literature review should explain why the question matters and how previous work informs the method. It should not become a detached textbook chapter. Materials and methods should be detailed enough to explain what was actually done and how the reported result was produced.

12. Formulate aims and hypotheses before collecting data

BIMA3240 explicitly expects students to set up hypotheses and research aims. Make these operational before data generation. A broad aim such as “study biomedical images with AI” is not yet a defensible research aim. Specify the modality or image type, biological or clinical target, dataset or experimental system, task, comparison and performance or scientific outcome. If the work is exploratory and a directional hypothesis is not appropriate, state clear research questions instead. The analysis should then be selected to answer those questions, not the other way around.

13. Match the method to the Biomedical Imaging environment

The current curriculum combines practical imaging, microscopy, bioimage informatics and statistical data analysis. Mandatory studies include Statistical Data Analysis, Bioimage Informatics 1 and 2, and Physical Basis of Medical Imaging, while selectable modules allow deeper work in medical imaging, cell biology and microscopy, or data analysis. These course environments show the programme’s methodological breadth, but they do not mean every thesis uses every technique. A microscopy thesis, PET project and machine-learning image-analysis thesis can all require different evidence and validation logic.

14. Treat an image as a measurement, not simply a picture

In biomedical imaging, an image is produced by an acquisition and processing chain. The modality, specimen or participant, instrument, acquisition settings, reconstruction or preprocessing and analysis choices can all affect the final measurement. Record parameters that materially influence the result. If acquisition settings change between groups or time points, explain why and how this affects comparability. If an image is transformed before measurement, document the transformation. Reproducibility requires more than saving the final figure that appears in the thesis.

15. Define the experimental unit before counting images

A common imaging mistake is treating every image, field of view, slice or segmented object as an independent experimental replicate. Multiple images may come from the same specimen, animal, participant or acquisition session. Define the experimental unit and the unit of analysis explicitly. Technical replication can improve measurement precision, but it does not automatically increase the number of independent biological units. Statistical analysis should respect this dependency, otherwise uncertainty may be understated and apparently strong significance may come from pseudoreplication.

16. Separate technical, biological and clinical evidence

A strong Biomedical Imaging thesis states exactly what level of evidence has been produced. A bench or phantom experiment can demonstrate technical performance. Cell culture or tissue imaging can support biological findings in that model. Animal or other in-vivo models add a different level of biological evidence. Retrospective human images can support bounded analytical findings. Prospective clinical utility, diagnosis, treatment benefit or patient outcome claims require evidence at those levels. Do not allow clinically meaningful vocabulary to make lower-level evidence sound stronger than it is.

17. Microscopy studies need acquisition and sampling discipline

For microscopy projects, describe sample preparation, staining or labels, microscope and objective where relevant, acquisition settings, field-selection rules and any image-processing steps that affect measurement. Avoid selecting only attractive fields for quantitative analysis. If several cells or fields come from one biological sample, preserve that hierarchy in analysis. If segmentation or object detection is used, explain how thresholds or algorithms were selected and how errors were checked. A visually convincing image is not itself evidence that a quantitative biological claim is reliable.

18. Medical-imaging studies must define the modality and reference

For MRI, CT, PET or other medical-imaging work, state the modality, acquisition or reconstruction context, target and validation reference. A model predicting a clinical label needs a clear account of where that label came from. Segmentation needs a reference segmentation or another justified validation strategy. Quantitative imaging needs a defined outcome and appropriate comparison. A result produced on one scanner, protocol or retrospective cohort should not automatically be generalised to all scanners, institutions or future patients.

19. Bioimage informatics needs a protected validation design

For machine learning or other data-driven image analysis, preserve training, tuning and test separation. The test set should not become another development set through repeated inspection and model adjustment. Document preprocessing, augmentation, feature extraction, architecture or algorithm version, hyperparameter selection and evaluation metrics that materially affect results. If images from the same participant or specimen appear in different data splits, leakage can make performance look stronger than it is. Split at the correct independent unit whenever the research question requires that independence.

20. Ground truth is also a measurement problem

Reference labels are not automatically perfect. Histopathology, expert annotations, clinical records, manual segmentations and instrument-based references each have their own uncertainty. If an annotation is used as ground truth, report who or what produced it, whether multiple annotators were involved, how disagreement was handled and whether the reference was independent of the method being tested. When uncertainty in the reference is important, discuss it explicitly rather than treating all apparent model errors as failures of the model alone.

21. Statistical analysis must follow the data structure

The current programme includes mandatory statistical-data-analysis training, but no statistical method becomes correct merely because it is taught in the curriculum. Define the outcome, independent units, groups or predictors, repeated measurements and missingness first. Then choose a method whose assumptions fit the design. Report analysed sample sizes and exclusions. Where appropriate, present effect magnitude and uncertainty rather than relying only on a p-value. A statistically significant difference can still be too small, unstable or context-specific to support a strong scientific conclusion.

22. Report failed acquisitions and exclusions

Biomedical imaging datasets often lose observations because of motion, low signal, corrupted files, failed staining, segmentation failure, scanner problems or protocol deviations. These losses can be scientifically important. Define exclusion rules before looking at outcomes where possible, and report how many samples or images were excluded and why. If exclusions differ across groups, assess whether this could bias the result. Quietly removing poor-quality data can make a method appear more robust than it is and can make the final analytic sample difficult to reconstruct.

23. Version the computational parts of the thesis

If software, scripts, firmware, reconstruction tools or machine-learning code contribute to results, keep enough version information to reproduce the analysis. Record relevant software versions, packages, parameters and model checkpoints or configuration files. Link raw or source data to preprocessing outputs and final figures through a clear naming or provenance system. A research notebook, repository or controlled project folder can help. Reproducibility is especially important when the final figure is several computational steps away from the original acquisition.

24. Ethics requirements depend on the actual project

Biomedical Imaging is a biomedical programme, but not every thesis follows the same ethical-review route. A microscopy project using established non-human material, a retrospective image-analysis project and an interventional study with people have different requirements. University of Turku distinguishes human-sciences ethical review and legally regulated medical research. The correct route must be determined from the actual project. Do not assume that the programme title itself either creates or removes an ethics requirement.

25. Medical research under the Act has a higher legal gate

University guidance states that medical research falling within the Medical Research Act, including covered research involving intervention in a person’s physical or psychological integrity, requires a favourable statement from the relevant regional medical ethics committee before research begins. This is not a paperwork step that can be repaired after data collection. If a thesis may fall within this boundary, the supervisor and research group should resolve the classification, responsible organisation and existing approvals before the student begins the relevant work.

26. Human-participant projects need consent, privacy and data planning

Where a thesis involves human participants but follows the human-sciences route, current University guidance can require participant information, informed-consent materials, recruitment description, a data management plan, a privacy notice and other supporting documents depending on the study. Ethical review is still conditional, not automatic for every interview or dataset. The practical rule is to screen the design early and document why a review, permit or particular consent route is or is not required before collecting research data.

27. A research permit does not replace ethical review or consent

University of Turku has a research-permit process for research concerning relevant University units, staff, students or organisational settings. Other hospitals, wellbeing services counties, research institutes or organisations may have their own permissions. A research permit authorises research within the organisation’s scope; it does not automatically provide participant consent, ethical approval or lawful access to personal data. For a collaborative Biomedical Imaging thesis, map each approval to the activity it actually authorises.

28. Human images and metadata can be personal data

Research images, identifiers, dates, clinical variables and linked metadata can contain or become personal data when a person is identifiable or can be re-linked. Plan storage, access, pseudonymisation or anonymisation and data transfer before analysis. Use only the data needed for the research question. Do not treat removal of a name as proof that an image dataset is anonymous. The research group and responsible researchers remain accountable for data administration under University data-protection policy and good scientific practice.

29. Animal work has its own conditional competence requirements

Some Biomedical Imaging projects may use animal models or regulated animal procedures, while many do not. University of Turku’s Central Animal Laboratory states that only trained and qualified persons may perform regulated procedures, care for or euthanise laboratory animals, design experiments or apply for licences as applicable. Function A/D training is specifically relevant to students and researchers who will be involved in procedures. Do not apply animal-study rules to projects that contain no regulated animal work, but do not begin animal procedures without the required competence and project approvals.

30. Keep protected data out of uncontrolled external tools

Imaging projects can contain patient images, unpublished laboratory results, collaborator data or proprietary methods. Do not upload protected material to external AI, cloud or analysis services merely because the tool is convenient. First establish that the service, data transfer and processing are allowed for the project. The same principle applies to generative AI used for coding, summarisation or drafting. Student responsibility for scientific accuracy, source use and confidentiality remains with the student even when a tool produces apparently plausible output.

31. The thesis is written in English and assessed by two examiners

The current programme description states that the master’s thesis is written in English. It also requires at least two examiners. The responsible professor appoints the examiners, and one can be the supervisor. BIMA3240 uses the 0–5 assessment scale. The programme-level description states that the head of the Department of Biomedicine decides on approval based on the examiners’ opinions. This formal programme-level wording should control any simplified interpretation of course-level wording about acceptance by the responsible professor.

32. Turnitin is a mandatory originality control for degree theses

University of Turku’s current Turnitin guidance states that originality checking is mandatory for theses that are part of a degree. For a master’s thesis, the Turnitin check is approved through the electronic UTUGradu process. Use Turnitin as an originality and source-use control, not as a proxy for scientific quality. A low similarity percentage does not prove that methods are valid, and a high percentage requires interpretation rather than an automatic misconduct conclusion. Correct quotation, citation and honest authorship remain fundamental.

33. UTUGradu handles the final electronic thesis process

UTUGradu is the University’s electronic process for higher academic degree theses. It combines the originality check with examination and approval, electronic publication and electronic archiving. Plan enough time for final revision, supervisor instructions, Turnitin and the UTUGradu workflow rather than treating upload as the moment the research ends. Check the current faculty and programme instructions at submission because administrative details can change even when BIMA3240’s core research route remains stable.

34. Run an evidence audit before final submission

Before freezing the manuscript, take every major conclusion and trace it backwards. What data support it? What is the independent experimental unit? What acquisition settings and preprocessing produced the measurement? What validation reference was used? Does the statistical model match the data structure? Are missing observations explained? Does the conclusion stay within the specimen, model, cohort, scanner, laboratory or dataset actually studied? This final audit is particularly important when a technically strong image result could easily be written as a broader biological or clinical claim.

35. Final Biomedical Imaging thesis checklist

Confirm programme 105701 and your current personal study plan; BIMA3240 at 40 ECTS; year-two scheduling; a suitable research-group project; topic discussion with the responsible professor; supervisor-accepted thesis plan; thesis-plan seminar; responsible-professor approval before practical work; applicable ethics, permit, privacy and animal requirements; clear experimental unit and imaging protocol; reproducible preprocessing and analysis; correct validation reference; appropriate statistics; transparent exclusions; protected-data controls; English scientific manuscript; final thesis seminar; two-examiner assessment; 0–5 grading; mandatory Turnitin; and final UTUGradu submission. Most importantly, keep every conclusion at the evidence level your project actually supports.