Biomedical Thesis Topics for Engineering Students: Ward Devices, Images and Screening Tools in Pakistan

Which hospital device, signal or screening tool can I test with data I can actually reach? Will my ethics committee read my study as device testing, record review or a bench experiment? What will I compare my result against: a reference method, a published cut-off or a regulator’s rule? Can one lab and one hospital carry the whole project in a single semester?

Those four questions decide whether a biomedical idea becomes a thesis or a stalled prototype. Twelve topics follow, and each starts from a Pakistani hospital record, a registered dataset or a published study. Biomedical here means biomedical engineering: ward devices, medical signals and images, screening tools and rehabilitation hardware. Laboratory diagnostics sit in medical laboratory technology research topics, and gene and microbe work sits in biotechnology thesis topics. The short biomedical list inside engineering thesis topics gives broad directions, while the topics here are narrower.

Where a Biomedical Engineering Thesis Fits in Pakistani Programmes

Thesis rules differ by university, so read your own handbook first. The table shows how three institutions describe their biomedical engineering routes on their own websites.

Institution	Route and thesis scale as described by the university
NED University, Karachi	M.Engg (Biomedical) and a PhD in Biomedical Engineering, both described as approved by HEC. Thesis options are listed in a university office order.
Ziauddin University, Karachi	MS Biomedical Engineering of 30 credit hours, taken either as coursework only or as 24 credit hours plus a 6 credit hour thesis. The thesis can start after 18 credit hours.
Liaquat University of Medical and Health Sciences	The Institute of Biomedical Engineering and Technology offers MS Biomedical Engineering and MPhil Biomedical Sciences. A 6 credit hour research option can replace two courses.

Plan the scale before the topic. A 6 credit hour thesis leaves little time for building anything new, so keep it to one device, one site and one measurement. Topics 2, 3, 4 and 8 suit that size. Topics 5, 6, 7 and 10 need more data handling, so they suit a larger MS thesis or a PhD chapter. The MPhil thesis guide explains how the supervisor and evaluation steps usually run.

Rules That Touch a Biomedical Thesis in Pakistan

Device rules from DRAP

The Medical Devices Rules, 2017 regulate devices under the DRAP Act, 2012, and the text published by DRAP is amended up to April 2022. The rules define a clinical investigation as a systematic study on people that assesses a device’s safety and performance. They sort devices into four classes by hazard, from Class A (low) to Class D (high), and they give a lung ventilator as a Class C example.

Several provisions matter to a student. A hospital must run a maintenance management system with an up-to-date inventory (rule 55), and anyone who maintains or calibrates a device must be trained to do so (rule 58). Importing a small quantity of a device for clinical investigation needs a permit (rules 22 and 23). A device meant for investigation must also be labelled for qualified investigators only (rule 43). So speak to your supervisor before any home-built device touches a patient.

Position correct as of November 2026, based on the DRAP text amended to April 2022. Check the DRAP medical devices pages for later amendments.

Ethics review and permissions

Studies on people usually need approval from your institution’s ethics review committee before data collection starts. The National Bioethics Committee for Research describes its role as reviewing research on human subjects, and it says studies without direct participants, such as chart review, may be exempted. Ask your supervisor whether a separate national step applies to your design.

Record-only work still needs a written permission letter from the hospital. A Peshawar CT dose study, for example, reports that its data were collected after an exemption from the institutional ethics committee.

Biomedical Thesis Topics With a Local Starting Point

Every topic below names a Pakistani setting, a comparison and a measurable result. Change the site or the device to suit your university, but keep those three anchors.

On the Ward and in the Hospital Store

Nursing students who want the practice side of ward devices can also look at medical-surgical nursing research topics.

1. Pulse oximeter readings across skin tones in one ICU

Pulse oximeters can read too high in people with darker skin, which hides low oxygen. A 2023 review of the evidence found decreased accuracy in dark skin. A student letter in the Journal of the Pakistan Medical Association argues that Pakistani hospitals should test their devices on local skin tones. However, it relies on foreign studies and reports no Pakistani measurements. That gap is the opening.

Pair each oximeter reading with the arterial blood gas saturation already drawn for care, and record skin tone with a colour chart. Report bias and limits of agreement. Then count readings of 92% or more when the blood value is below 88%, the hidden hypoxaemia definition used in the letter. This suits MS level. Record low perfusion for every patient, because it can distort readings.

2. Phototherapy lamp output in the newborn units of one city

Newborn jaundice is treated with blue light, and the dose depends on lamp output at the baby’s skin. In 17 hospitals on Java, Indonesia, irradiance fell below the standard level in nine hospitals and below the intensive level in eight. In Kaduna, Nigeria, only 13.3% of facilities had devices that reached the conventional level at their usual distance.

Repeat that design in one Pakistani city. Measure irradiance with a calibrated radiometer at five points across the mattress, as a Dutch study of six LED devices did, and compare the lit area with a newborn’s body size. The unit of analysis is the lamp, not the reading, so 30 lamps measured at five points give 30 devices and not 150. Confirm radiometer access before you propose it. This fits a final-year project or an MS thesis.

3. Why ventilators stay broken: repair times in one hospital’s maintenance log

In February 2026 the Express Tribune reported 431 non-functional ICU ventilators in Punjab government hospitals. Of these, 191 awaited repair, 185 were beyond repair and 55 were condemned. The report also named poor after-sales support and missing trained operators among the causes for pandemic-era machines. Officials said 300 more ventilators had been approved. Those figures are as reported in February 2026, so check for later health department data.

A thesis can work one level below the headline. Take one hospital’s work orders for ventilators and anaesthesia machines, then code each fault by cause, part, supplier and days to repair. Curves of repair time by cause would show where the waiting happens. A 2025 teaching case on equipment maintenance across 36 Punjab districts frames the same problem as maintenance network design. No patient data are involved, but you need a letter from the medical superintendent. Audit the log first, because missing dates will sink the analysis.

4. Do the pumps and oximeters on one ward carry a DRAP registration number?

The DRAP rules expect a device label to carry the enlistment or registration number (rule 42). They also expect a hospital’s maintenance system to cover the purchase of enlisted or registered devices (rule 55). Test that in practice. List every infusion pump or pulse oximeter on one ward, photograph each label, and look the number up in the DRAP lists of enlisted and registered devices. Report the share that match, the share with a label but no match, and the share with no readable label.

Older stock complicates the count. The exemption table in rule 52 ran out between March and December 2022, depending on class, so a pump bought earlier may be legitimately unlisted. Record the purchase year for every device. Anonymise the ward and hospital in the thesis, and agree the permission letter first, because findings can embarrass a facility. This suits an MS thesis in biomedical engineering or health technology management.

Signals and Images From Pakistani Patients

5. Setting a chest X-ray TB triage threshold at a public chest clinic

Computer-aided detection software scores a chest X-ray, so clinics can send only high scorers for Xpert testing. A retrospective study at two private centres in Karachi, run from 2013 to 2015, reported an area under the curve of 0.79 for the score alone and 0.83 with age and sex added. Later work tested version 6 of the software on 5,565 people in Karachi, and a separate study tested 694 people with diabetes.

All of those data came from private or screening settings in one city. Ask whether the same cut-off holds at a public chest clinic elsewhere. Report sensitivity and specificity against Xpert, then show how many Xpert tests each cut-off would save. Check software licences first, because commercial tools may not allow student use. The topic suits MS level, and it needs Xpert-linked records, so confirm the clinic keeps them.

6. Testing a diabetic retinopathy classifier on fundus images from one local clinic

Most published classifiers train on public image sets. The RDS-DR paper describes a new collection called DR-Insight, gathered from Pakistani eye hospitals and online sources, and trains on 9,860 images in total. A PCSIR registry entry plans a graded fundus dataset of up to 50,000 images at Al Ibrahim Eye Hospital in Karachi, with three retina specialists grading each image. Ask the sponsor whether any part is shared for student use, and do not assume it is.

Narrow the thesis to one question: how much accuracy does a model trained on a public set lose on images from one clinic’s camera? Report results per disease grade and per image-quality band, not as one overall figure. You need an agreement with the clinic and a qualified grader. Computer engineering students can pair this with the model ideas in computer science thesis topics, and a dental imaging version of the same question sits in dentistry thesis topics.

7. How much does scan quality hurt ECG arrhythmia classification?

A 2024 paper in Sensors turned ECG plot images from patients at Pakistani healthcare institutes into time series before applying deep learning. Image quality is the obvious weak point of that workflow, and a thesis can test it directly.

Start with clean images, then degrade them in steps: blur, skew, low resolution and phone-photo glare. Plot how the score for each rhythm class falls at each step. This is a bench study with no new patient data, so ethics review is light, although dataset terms still apply. Report per-class results, because rare rhythms hide inside an overall average. The topic suits MS in electrical or biomedical engineering.

Screening Tools and Radiation Doses for Children

8. A smartphone earbud hearing probe against a hospital device in a noisy postnatal ward

Newborn hearing screening relies on otoacoustic emission devices. A University of Washington team built a probe from off-the-shelf earphones and a microphone for about US$10 in materials. In 201 ears at three clinics, it detected hearing loss with 100% sensitivity and 88.9% specificity, and the authors released their design as open source. A Karachi study of 120 newborns at Creek General Hospital found the standard emission test practicable.

Noise is the practical problem, because ward noise raises false refers. A bench thesis can build the probe, test it in an ear coupler, and measure how background noise in a postnatal ward changes pass or refer results in adult volunteers. Move to newborns only if the ethics committee and the hospital approve, since an unregistered device on babies falls under the investigational rules described above. This suits MS level.

9. Can Lady Health Workers record usable lung sounds from children?

Researchers in Bangladesh asked community health providers to record lung sounds from 990 children aged 2 to 59 months at four chest positions. Their Edinburgh PhD thesis concluded that the providers could record good-quality sounds without a large increase in workload. A larger randomised trial of an automated classifier is now running there.

A Pakistani thesis can borrow only the first step. Compare recording quality between Lady Health Workers and clinic doctors, scored against quality thresholds set by a paediatrician panel. Leave out the classifier, since it needs far more data than one thesis can collect. Consent must come from carers, so plan guardian permission early. Record the setting too, because a quiet clinic room does not predict a noisy courtyard. The topic suits MS in biomedical engineering or public health.

10. Paediatric CT head dose by weight band at one children’s hospital

A Lady Reading Hospital study in Peshawar set institutional diagnostic reference levels from 1,001 adult CT scans on one 160-slice scanner over three months. It states its own limits plainly: no paediatric levels, and no patient height or weight. The authors hope other hospitals will gather their own dose data. The regulatory text to cite is the Pakistan Nuclear Regulatory Authority’s Regulations on Radiation Protection, PAK/904 Rev.1, which define the diagnostic reference level.

Take the missing piece: one children’s hospital, head CT, and patients grouped by weight. A paediatric dose study applying European guidance used four weight groups: under 15 kg, 15 to under 30 kg, 30 to 50 kg and over 50 kg. Export the volume dose index and the dose length product from the console, then report the median and the 75th percentile for each group. Retrospective dose data may qualify for an ethics exemption, as at Peshawar, but ask first. Check that weight is actually recorded before you commit. This suits MS in medical physics or biomedical engineering.

Limbs and Rehabilitation

11. Wear and skin comfort of a low-cost prosthetic liner through a hot summer

A Peshawar rehabilitation centre has registered a study of a locally made liner for people with below-knee amputation, listing socket comfort and six-minute walk distance as outcomes. The registry entry shows a very small enrolment and a one-year window. A longer follow-up through the hot months would add what that study cannot show. Measure skin temperature, sweat complaints and liner wear against the user’s usual silicone liner.

Custom devices raise a rule. The DRAP rules allow a custom-made device to be fabricated only by a qualified practitioner for a patient under their care (rule 53). Confirm with the centre and your supervisor who may fabricate each liner before you plan any fitting. The student then designs the measurements, and the clinical team handles the fitting. This suits MS in biomedical engineering with a rehabilitation partner.

12. Trunk motion and low back pain in amputees: adding a wearable sensor to a questionnaire study

At the Pakistan Institute of Prosthetic and Orthotic Sciences in Peshawar, a 2020 cross-sectional study of 264 people with unilateral lower-limb amputation measured back pain and function with questionnaires only. Other studies at the same centre have used its gait laboratory with a motion-analysis camera system.

A thesis could add one objective measure. Record trunk motion with a body-worn inertial sensor during a walking test, compare people with and without back pain, and check the result against the questionnaire score. Keep the sample modest and report effect sizes instead of a verdict. Prosthetic type, stump length and walking speed all change trunk motion, so record each one and adjust for it. Access depends on the centre’s permission and on each participant’s consent.

Five Steps to Narrow a Device Idea Into a Thesis

1. Name one device, signal or image set. “AI in healthcare” is a field. “A retinopathy classifier tested on one clinic’s fundus camera” is a project.
2. Name one site and one population. One ICU, one city’s newborn units or one prosthetic centre keeps permissions manageable.
3. Fix the comparator before collecting data. It may be a reference method, a published cut-off or a regulator’s rule. The guide on choosing a research methodology helps you match the design to that comparator.
4. Count the right unit. Lamps, patients, images and clinics are different units, and your sample size depends on which one you analyse.
5. List what you cannot measure. Write down the missing weights, the unlisted devices or the closed lab, together with the permission letters you still need.

Common Mistakes in Biomedical Thesis Proposals

Starting From the Algorithm

A neural network is a method, not a question. First name the clinical or engineering problem, and then choose the model that fits the data you can reach.

Treating a Foreign Study as Local Evidence

The Java lamp audit says nothing about lamps in Lahore. Use foreign work to borrow a design, and then collect your own Pakistani measurements.

Building Before Asking Permission

A prototype that cannot legally or ethically reach a patient has no clinical result to report. Settle the ethics route and the device rules while you are still writing the proposal.

Leaving Out the Comparator

A new device with no reference method gives numbers without meaning. Name the comparison in the first paragraph of the proposal.

Frequently Asked Questions About Biomedical Thesis Topics

Can I build my own device for a biomedical engineering thesis and test it on patients?

The safer route is to test on the bench first, using couplers, phantoms or volunteers where your committee allows it. Any test on patients needs ethics approval, and the DRAP rules treat devices used for investigation separately, including an import permit for small quantities and a special label. Your supervisor and clinical partner should decide how far the project goes. This is general information, not legal advice.

How is a biomedical thesis different from a medical laboratory or clinical research thesis?

A biomedical engineering thesis tests how a device, signal or tool performs. A medical laboratory thesis tests a diagnostic method on samples, and a biotechnology thesis works with genes, cells or microbes. Clinical questions about patients and treatments sit in medical research topics. If your idea mixes two of these, choose the field where your supervisor can examine the methods.

Do I need patient data for a biomedical thesis?

Not always. Topics 3, 4 and 7 use maintenance logs, device labels and image files, and none involves patient records. You still need written permission to use hospital logs or ward devices. To see how earlier students structured similar work, browse the HEC research repository for biomedical theses.

Choosing Among the Twelve

Pick the topic whose permission letter you can get fastest, because access decides most biomedical theses more than ideas do. Once one topic is shortlisted, the research proposal guide shows how to turn it into a proposal. Device audits such as topics 2 and 3 can also become short papers, and turning a thesis into a journal paper explains how. Other subject lists are gathered in the thesis topics hub.

If you have settled on a topic and want a proposal outline, a method check or an editing pass, academic support for your biomedical project can be arranged on request. Any work received is for reference and learning and should be used according to your institution’s policy.