Advanced Biomedical Engineering interview questions

125 questions with answers (BME-301 to BME-425). Try each one aloud before reading the answer.

Anatomy & Physiology for Engineers

BME-301 · Use the GHK equation to find the resting potential for P_K : P_Na : P_Cl = 1 : 0.04 : 0.45 with K⁺ 5/140, Na⁺ 145/15 and Cl⁻ 110/10 mM (out/in) at 37 °C.

Numerator = 5 + 0.04×145 + 0.45×10 = 15.3. Denominator = 140 + 0.04×15 + 0.45×110 = 190.1. V = 61.5 log₁₀(15.3/190.1) = 61.5 × (−1.094) ≈ −67 mV.

BME-302 · Model the arterial system with a two-element Windkessel. What is the diastolic decay time constant for R = 1 mmHg·s/mL and C = 1.5 mL/mmHg?

C·dP/dt + P/R = Q(t). In diastole Q = 0, so P(t) = P₀e^(−t/RC) with τ = RC = 1.5 s. Stiffer arteries (lower C) give faster decay and a wider pulse pressure.

BME-303 · How does cable theory explain why larger axons conduct faster?

The length constant is λ = √(r_m/r_i). Internal resistance per unit length falls as 1/d², while membrane resistance per unit length falls only as 1/d, so λ grows with √d. Current spreads farther ahead, and myelin further raises r_m and lowers capacitance.

BME-304 · How does raising [K⁺]o from 5 to 8 mM affect E_K and cardiac excitability?

E_K = 61.5 log₁₀(8/140) ≈ −76 mV, compared with −89 mV, so the membrane depolarizes. This first brings it closer to threshold, then inactivates Na⁺ channels, slowing conduction. The result is peaked T waves, widened QRS and a risk of arrhythmia or asystole.

BME-305 · Calculate oxygen delivery for CO = 5 L/min, Hb = 15 g/dL, SaO₂ = 98% and PaO₂ = 100 mmHg.

CaO₂ = 1.34×15×0.98 + 0.003×100 ≈ 20.0 mL/dL = 200 mL/L. DO₂ = CO × CaO₂ = 5 × 200 = 1000 mL O₂/min.

BME-306 · Use the Fick principle to calculate cardiac output for VO₂ = 250 mL/min, CaO₂ = 200 mL/L and CvO₂ = 150 mL/L.

CO = VO₂/(CaO₂ − CvO₂) = 250/(200 − 150) = 5 L/min.

BME-307 · Calculate inulin clearance for urine concentration 25 mg/mL, urine flow 1.2 mL/min and plasma concentration 0.25 mg/mL.

C = U·V/P = 25 × 1.2 / 0.25 = 120 mL/min, which approximates GFR because inulin is freely filtered and neither reabsorbed nor secreted.

BME-308 · What does a ventricular pressure–volume loop tell an engineer?

The loop area is the stroke work and its width is the stroke volume. The slope of the end-systolic pressure–volume relation (Ees) indexes contractility independently of loading. The end-diastolic curve reflects chamber stiffness. Such loops are used to evaluate VADs and heart-failure therapies.

BME-309 · For a ventilated patient with Vt = 500 mL, C = 50 mL/cmH₂O, R = 10 cmH₂O/(L/s), inspiratory flow 1 L/s and PEEP 5 cmH₂O, find the peak and plateau pressures.

Equation of motion: P = Vt/C + R·V̇ + PEEP. Plateau = 500/50 + 5 = 15 cmH₂O. Peak = 15 + 10 × 1 = 25 cmH₂O.

BME-310 · How does the strength–duration curve guide stimulator pulse-width design?

Lapicque: I = I_rh(1 + t_c/t), with rheobase I_rh and chronaxie t_c. The energy per pulse, ∝ I²t, is minimized when the pulse width equals the chronaxie. Stimulus charge falls as the pulse gets shorter. Pacing pulses of about 0.4–0.5 ms trade off battery energy against safety margin.

Biomedical Instrumentation

BME-311 · Outline the design of a diagnostic ECG front end.

Defibrillation and ESD protection, then an instrumentation amplifier with input impedance above 10 MΩ, CMRR above 100 dB and modest gain (about 10) to tolerate ±300 mV electrode offset. Then a 0.05 Hz high-pass filter, further gain to about 1000, anti-aliasing at about 150 Hz, a driven right leg, galvanic isolation and an ADC.

BME-312 · A 20 mV, 50 Hz common-mode voltage is present, electrode impedances differ by 20 kΩ and the common-mode input impedance is 100 MΩ. What differential error results?

V_d ≈ V_cm × ΔZ/Z_cm = 20 mV × 20×10³/100×10⁶ = 4 µV. This error is created before the amplifier, so CMRR cannot remove it. Higher Z_cm and good skin preparation reduce it.

BME-313 · An amplifier has Ad = 1000 and CMRR = 100 dB, and 1 V of common-mode voltage is present. What is the error relative to a 1 mV ECG?

Ac = Ad/10⁵ = 0.01, so the common-mode output is 10 mV, equivalent to 10 µV at the input. That is 1% of a 1 mV ECG, or −40 dB.

BME-314 · How do you choose the AC-coupling corner after the first ECG stage?

Electrode offsets up to ±300 mV would saturate a high-gain stage, so a high-pass filter removes DC. For 0.05 Hz, RC = 1/(2π×0.05) ≈ 3.2 s, for example 1 µF with 3.2 MΩ. The long time constant needs a fast-restore circuit to recover after saturation from defibrillation or lead-off.

BME-315 · How is a patient monitor protected against defibrillation pulses?

Series current-limiting resistors in each lead, voltage clamps (gas discharge tubes, TVS diodes or diode clamps to the rails) and adequate creepage. The design must survive several kV, limit energy diverted from the patient, and recover the trace within a few seconds.

BME-316 · How is a pulse oximeter calibrated, and why can't it be calibrated from first principles?

Tissue scattering and path-length differences make the Beer–Lambert law inaccurate. Manufacturers build an empirical R-to-SpO₂ curve from controlled desaturation studies on volunteers against arterial CO-oximetry. A commonly quoted approximation is SpO₂ ≈ 110 − 25R.

BME-317 · What are the main error sources in pulse oximetry?

Motion artifact, low perfusion, ambient light, venous pulsation, nail polish, dyes such as methylene blue, carboxyhaemoglobin (reads falsely high), methaemoglobin (pushes the reading toward about 85%), and overestimation in darker skin pigmentation, which can hide occult hypoxaemia.

BME-318 · A disposable blood pressure transducer has a sensitivity of 5 µV/V/mmHg with 5 V excitation. What is the output at 100 mmHg, and how is the system checked?

Output = 5 µV/V/mmHg × 5 V × 100 mmHg = 2.5 mV. The system is zeroed at the phlebostatic axis and checked with a fast-flush square-wave test to assess natural frequency and damping of the catheter–tubing system.

BME-319 · How does a fluid-filled catheter system distort arterial pressure waveforms?

It behaves as a second-order system. Compliance from bubbles or soft tubing, and long narrow tubing, lower the natural frequency. An underdamped system overshoots and overestimates systolic pressure, and an overdamped one underestimates it. A high natural frequency with damping ratio of about 0.6–0.7 is best.

BME-320 · Calculate the thermal noise of a 10 kΩ electrode resistance at 37 °C over a 100 Hz bandwidth.

v_n = √(4kTRB) = √(4 × 1.38×10⁻²³ × 310 × 10⁴ × 100) ≈ 1.3×10⁻⁷ V, or about 0.13 µV rms. This is negligible compared with amplifier and electrode noise.

BME-321 · Strong 50 Hz noise appears on one patient's ECG but not on others connected to the same monitor model. How do you troubleshoot?

Suspect patient-specific causes: dried or poorly applied electrodes, unequal impedances, a loose RL or reference lead, or damaged cable shielding. Also check proximity to unshielded mains cords or a bed motor. Re-prep the skin and replace electrodes and leads before relying on the notch filter.

Biosignal Processing (ECG, EEG, EMG)

BME-322 · What causes motion artifacts in biopotential recordings, and how are they reduced?

Electrode movement disturbs the electrode–electrolyte double layer, and skin stretch changes skin potential. Reduce them with skin abrasion, Ag/AgCl gel electrodes, secure cable strain relief, adaptive filtering, and accelerometer- or impedance-referenced cancellation.

BME-323 · How would you design an adaptive canceller for powerline interference?

Use sine and cosine reference signals at the mains frequency, weighted by two LMS coefficients: w(n+1) = w(n) + 2µe(n)x(n). The output e(n) = ECG − estimate. This tracks amplitude, phase and slight frequency drift and gives a very narrow notch with little QRS distortion.

BME-324 · Design a second-order digital notch at 50 Hz for fs = 500 Hz.

ω₀ = 2π×50/500 = 0.2π and cos ω₀ ≈ 0.809. H(z) = (1 − 1.618z⁻¹ + z⁻²)/(1 − 1.618r z⁻¹ + r²z⁻²), with r ≈ 0.95–0.99. Poles closer to the unit circle narrow the notch but lengthen the ringing.

BME-325 · Why shouldn't a diagnostic ECG use a simple 0.5 Hz causal high-pass filter?

Non-linear phase and the low-frequency cutoff distort the ST segment and can mimic ischaemia. Guidelines recommend 0.05 Hz for causal filters, or up to about 0.67 Hz with linear-phase or zero-phase filtering.

BME-326 · How can the fetal ECG be extracted from abdominal recordings?

The maternal ECG is roughly 5–10× larger and overlaps in frequency. Use maternal template subtraction, adaptive cancellation with a maternal chest reference, or blind source separation with ICA on multichannel recordings. Then detect fetal QRS complexes to obtain fetal heart rate.

BME-327 · How are ocular and muscle artifacts removed from EEG?

Remove eye blinks and movements by regression on EOG channels or with ICA, identifying and removing artifact components before back-projection. Reduce muscle artifact with ICA, canonical correlation analysis or band limiting. Reject heavily contaminated epochs.

BME-328 · Outline an ECG arrhythmia-detection pipeline and how you would validate it.

Filter, detect QRS, delineate waves, extract features (RR intervals, QRS width, morphology), then classify with rules or ML. Validate on annotated databases such as MIT-BIH using AAMI EC57 beat classes and inter-patient splits. Report sensitivity and positive predictivity for each class.

BME-329 · An evoked potential has an SNR of −10 dB and you need +20 dB. How many sweeps are needed?

Power SNR improves by a factor of N with averaging. The required gain is 30 dB, so 10 log₁₀N = 30 and N = 1000 sweeps, assuming stationary signal and uncorrelated noise.

BME-330 · How can respiration be derived from a single-lead ECG?

ECG-derived respiration uses three effects: QRS amplitude modulation as the cardiac axis shifts and thoracic impedance changes, respiratory sinus arrhythmia in RR intervals, and baseline wander. Fusing these estimates gives respiratory rate without a dedicated sensor.

BME-331 · What are the pitfalls of spectral HRV analysis?

RR series are unevenly sampled, so they need interpolation (for example at 4 Hz) or a Lomb–Scargle periodogram. Ectopic beats and artifacts must be edited. Recordings must be stationary, typically at least 5 minutes for short-term analysis. Breathing rate can shift HF power into the LF band.

BME-332 · Design a pattern-recognition controller for a myoelectric prosthesis.

Use multi-channel surface EMG with 150–250 ms analysis windows. Extract time-domain features: mean absolute value, waveform length, zero crossings and slope sign changes. Classify with LDA or a small neural network, keep total latency below about 300 ms, and retrain to handle electrode shift and fatigue.

Medical Imaging

BME-333 · What is the principle of PET?

A positron-emitting tracer such as ¹⁸F-FDG (half-life about 110 min) emits positrons that annihilate with electrons. Each annihilation produces two 511 keV photons travelling about 180° apart. Coincidence detection defines lines of response, which are reconstructed into quantitative maps of metabolic activity.

BME-334 · A 10 mT/m gradient is applied. What is the frequency difference between protons 1 mm apart?

Δf = (γ/2π) × G × Δx = 42.58 MHz/T × 0.01 T/m × 0.001 m ≈ 426 Hz. This frequency encoding maps position into the signal spectrum.

BME-335 · What is the axial resolution of a 5 MHz transducer with a two-cycle pulse?

λ = c/f = 1540/5×10⁶ ≈ 0.31 mm. Spatial pulse length = 2λ ≈ 0.62 mm. Axial resolution = SPL/2 ≈ 0.31 mm.

BME-336 · What round-trip attenuation does a 5 MHz beam suffer imaging at 8 cm depth with 0.5 dB/cm/MHz?

The path length is 16 cm, so attenuation = 0.5 × 5 × 16 = 40 dB. Time-gain compensation increases receiver gain with echo time to equalize brightness at depth.

BME-337 · How are noise and dose related in CT, and how is dose reduced?

Noise σ ∝ 1/√(mAs), so halving noise needs four times the dose. Dose is reduced with automatic tube-current modulation, lower kV for small patients or iodine studies, iterative or deep-learning reconstruction, tighter scan range and organ-based modulation.

BME-338 · What is the Rose criterion?

A lesion is reliably detectable when its contrast-to-noise ratio is about 3–5 or higher. It links required photon counts, and therefore dose, to lesion size and contrast in quantum-limited imaging.

BME-339 · Why is RF heating a greater concern at 3 T than at 1.5 T?

SAR scales approximately with B₀², so the same sequence deposits about four times more RF power at 3 T. Wavelength effects also cause B₁ inhomogeneity. Sequences are modified with lower flip angles, longer TR or parallel transmission to stay within SAR limits.

BME-340 · If you halve the pixel size in both in-plane directions on MRI, what happens to SNR, and how can you recover it?

Voxel volume falls 4×, so SNR falls 4×. SNR ∝ √(number of averages), so recovering it by averaging alone needs 16× the scan time. Alternatives are higher field, better coils or lower receiver bandwidth.

BME-341 · What does dual-energy CT offer?

Acquiring at two energy spectra exploits differences in photoelectric and Compton attenuation for material decomposition. This gives iodine maps, virtual non-contrast images, kidney stone composition, bone removal and reduced beam-hardening and metal artifacts.

BME-342 · What is SUV in PET, and what confounds it?

Standardized uptake value = tissue activity concentration / (injected activity / body weight). It is confounded by uptake time, blood glucose, residual activity in the syringe, the partial-volume effect in small lesions, attenuation-correction errors, reconstruction settings and body composition.

BME-343 · Which imaging would you choose first for suspected acute stroke, and why?

Non-contrast CT first. It is fast, widely available and reliably excludes haemorrhage before thrombolysis. Then CT angiography and perfusion to find large-vessel occlusion. MRI diffusion-weighted imaging is more sensitive for early ischaemia but is slower and less accessible acutely.

Biomaterials

BME-344 · How would you choose materials for a cementless femoral stem and bearing?

Ti-6Al-4V for the stem: its lower modulus reduces stress shielding and it osseointegrates well, often with a porous or HA coating. It wears poorly, so the head should be CoCr or ceramic articulating against highly cross-linked UHMWPE. Fatigue strength must be verified at the neck.

BME-345 · A surgeon wants to fix a titanium plate with stainless steel screws. What is your advice?

Avoid it. Dissimilar metals in electrolyte, combined with crevices and micromotion, increase the risk of galvanic and fretting-crevice corrosion, releasing ions and weakening fixation. Use screws of the same alloy family as the plate.

BME-346 · What are the key design factors for a bioresorbable coronary scaffold?

Radial strength during vessel healing (months), with degradation afterward. Polymers such as PLLA need thicker struts, which increases thrombosis risk and reduces deliverability. Magnesium alloys corrode faster and evolve hydrogen. Recoil, radiopacity markers and a predictable strength-loss profile are also key.

BME-347 · Estimate the bounds on modulus for a PEEK composite with 40 vol% HA (E_HA = 100 GPa, E_PEEK = 3.6 GPa).

Upper bound (Voigt, isostrain) = 0.4×100 + 0.6×3.6 ≈ 42 GPa. Lower bound (Reuss, isostress) = 1/(0.4/100 + 0.6/3.6) ≈ 5.9 GPa. Particulate composites lie near the lower bound.

BME-348 · Outline the biological evaluation of a new long-term implant under ISO 10993-1.

Prepare a biological evaluation plan within risk management. Start with chemical characterization (ISO 10993-18) and a toxicological risk assessment. Address cytotoxicity, sensitization, irritation, acute to chronic systemic toxicity, genotoxicity, implantation and carcinogenicity. Use existing data where justified rather than defaulting to animal tests.

BME-349 · How can a blood-contacting surface be made less thrombogenic?

Immobilize heparin. Graft protein-repellent PEG or zwitterionic and phosphorylcholine coatings. Use nitric-oxide–releasing polymers, smooth surface finishes, or promote endothelialization. Each approach must survive sterilization and long-term flow.

BME-350 · A matrix device follows Higuchi kinetics and releases 30% of its drug in 1 day. Estimate release at 4 days.

Q = k√t, so Q(4) = 30% × √4 = 60%. The Higuchi model is reasonable up to about 60% release, after which depletion causes deviation.

BME-351 · Calculate the accelerated aging time to simulate 1 year of shelf life at 55 °C, with ambient 25 °C and Q₁₀ = 2.

AAF = Q₁₀^((55 − 25)/10) = 2³ = 8. Time = 365/8 ≈ 46 days. Real-time aging must later confirm the result, and the aging temperature must avoid material transitions.

BME-352 · How do you design an orthopaedic implant against fatigue failure?

Use S–N data with high safety factors. A hip may see about 1–2 million gait cycles per year, so components are tested to about 10⁷ cycles. Minimize stress concentrations and notches, account for the reduction in fatigue strength from porous coatings, and use surface treatments such as shot peening.

BME-353 · Patients with metal-on-metal hips show raised blood cobalt and pseudotumours. What is the likely cause and remedy?

Edge loading from cup malposition, plus taper corrosion, produces large amounts of CoCr wear debris and ions, causing adverse reactions to metal debris. Remedies are monitoring, revision to ceramic or CoCr on highly cross-linked UHMWPE bearings, and avoiding MoM designs.

Biomechanics

BME-354 · Estimate hip joint force in single-leg stance for a 700 N person, with abductor moment arm 5 cm and body-weight moment arm 10 cm.

Supported weight ≈ 5/6 × 700 ≈ 583 N. Abductor force = 583 × 10/5 ≈ 1170 N. Treating the forces as vertical, joint reaction ≈ 1170 + 583 ≈ 1750 N, or about 2.5× body weight. A cane in the opposite hand reduces this.

BME-355 · What trade-offs shape total knee replacement design?

Conformity versus constraint. Conforming surfaces lower polyethylene contact stress but transfer more torque to fixation. Designers choose between retaining and substituting the PCL to control femoral rollback, and between fixed and mobile bearings. Patellar tracking, ligament balance and wear of the insert are also key.

BME-356 · A Kelvin–Voigt tissue (E = 10 MPa, η = 100 MPa·s) is loaded with σ = 1 MPa. What is the strain at 10 s?

ε(t) = (σ/E)(1 − e^(−tE/η)) with τ = η/E = 10 s. ε(10) = 0.1 × (1 − e⁻¹) ≈ 0.063.

BME-357 · How can stress shielding around a hip stem be predicted and minimized?

Use FE models with strain-energy-density remodelling rules to predict bone loss. Reduce it with lower-stiffness materials or hollow and tapered geometry, proximal-only porous coating to load the proximal femur, and shorter stems.

BME-358 · How do you build a subject-specific finite-element model of bone from CT?

Segment the bone and mesh it. Calibrate HU to density with a phantom and map density to modulus with a power law E = aρᵇ (b ≈ 2–3). Apply physiological loads and boundary conditions, check mesh convergence, and validate against cadaveric strain-gauge or DIC tests.

BME-359 · Calculate pulse wave velocity for E = 0.5 MPa, h = 1 mm, r = 10 mm and ρ = 1060 kg/m³.

c = √(Eh/(2ρr)) = √(0.5×10⁶ × 0.001 / (2 × 1060 × 0.01)) = √(500/21.2) ≈ 4.9 m/s.

BME-360 · Calculate the Womersley number in the aorta (r = 12.5 mm, HR = 72 bpm, ρ = 1060 kg/m³, μ = 3.5 mPa·s) and interpret it.

ω = 2π × 1.2 ≈ 7.54 rad/s. α = r√(ωρ/μ) = 0.0125 × √(7.54 × 1060/0.0035) ≈ 19. With α ≫ 1, inertia dominates, the velocity profile is blunt and phase-lagged, and Poiseuille's parabolic profile is invalid.

BME-361 · What mechanical requirements drive coronary stent design?

High radial strength with low recoil, minimal foreshortening, flexibility for delivery, thin struts and good conformability. Fatigue resistance over about 400 million cardiac cycles (10 years) is required. These are verified by FE analysis and accelerated pulsatile durability testing.

BME-362 · Estimate L5/S1 compression when lifting 200 N held 40 cm from the joint, with upper body 450 N at 20 cm and an extensor moment arm of 5 cm.

Moment = 200×0.4 + 450×0.2 = 170 N·m. Extensor force = 170/0.05 = 3400 N. Compression ≈ 3400 + 650 ≈ 4050 N, which exceeds the NIOSH 3.4 kN action limit.

BME-363 · What must a transtibial prosthesis design balance?

Socket fit and pressure distribution, avoiding pressure on bony prominences. Alignment for a smooth rollover. Foot stiffness and energy storage and return for efficient push-off. Weight. Durability. Users should be assessed with gait symmetry, metabolic cost and skin outcomes.

Therapeutic & Life-Support Devices

BME-364 · How does hemodialysis remove waste and fluid?

Blood flows past dialysate across a semi-permeable membrane. Small solutes such as urea, creatinine and K⁺ diffuse down concentration gradients. A transmembrane pressure drives ultrafiltration of excess water. Convection in haemodiafiltration adds clearance of middle molecules.

BME-365 · What are the main design challenges for implantable devices?

Biocompatibility and biostability, power (primary battery, rechargeable or wireless), hermetic sealing against moisture, miniaturization, MRI conditionality, EMI immunity, long-term reliability over years, secure wireless telemetry and safe explant or replacement.

BME-366 · Estimate pacemaker longevity for 2.5 V, 0.4 ms pulses into 500 Ω at 60 bpm with 100% pacing, 8 µA housekeeping current and a 1.2 Ah battery.

Pulse current = 5 mA. Charge per pulse = 2 µC, so the average pacing current is 2 µA. Total current = 10 µA. Life ≈ 1.2 Ah/10 µA = 120,000 h ≈ 13.7 years, ignoring output-circuit inefficiency.

BME-367 · A 100 µF defibrillator capacitor discharges into a 50 Ω transthoracic impedance. What is the time constant, and why does biphasic waveform design compensate for impedance?

τ = RC = 50 × 100×10⁻⁶ = 5 ms. Delivered current and duration vary with impedance. Impedance-compensating biphasic truncated exponential waveforms adjust pulse duration or tilt so that effective current is maintained across patients.

BME-368 · What are the key design issues for a closed-loop artificial pancreas?

CGM lag and noise, and slow subcutaneous insulin action that peaks around an hour. The algorithm (MPC or PID with insulin-on-board) must be conservative to avoid hypoglycaemia. Meal and exercise disturbances need handling. Sensor or pump faults need fail-safes, and pump, CGM and controller must interoperate.

BME-369 · A ventilated patient's peak pressure rises sharply but plateau pressure is unchanged. What is the cause?

Raised peak pressure with a normal plateau means airway resistance has increased: secretions, a kinked or bitten tube, or bronchospasm. If both peak and plateau rise, compliance has fallen: pneumothorax, mainstem intubation, pulmonary oedema or ARDS.

BME-370 · Heparin 25,000 units in 250 mL is ordered at 1000 units/h. What rate is correct, and how do pumps prevent 10× errors?

Concentration = 100 units/mL, so the rate is 10 mL/h. Entering 100 mL/h gives a 10× overdose. Dose error reduction software with drug libraries and hard and soft limits, barcode scanning and EHR–pump interoperability catch such errors.

BME-371 · Electrosurgical current of 0.5 A flows through a return pad. Compare heating with full contact of 150 cm² and partial contact of 15 cm².

Current density rises from 3.3 to 33 mA/cm², a 10× increase. Heating per unit volume ∝ J², so it rises about 100×, which can cause burns. This is why return electrode monitoring and proper pad placement are essential.

BME-372 · What are the key engineering challenges in LVAD design?

Minimizing haemolysis and thrombosis through blade and gap geometry and magnetic levitation of the rotor. Managing reduced pulsatility, which is linked to bleeding. Driveline infection and the need for transcutaneous power. Speed control to avoid suction events. Long-term reliability.

BME-373 · How does a medical linear accelerator deliver conformal radiotherapy?

Electrons accelerated by microwaves strike a target to produce MV photons. These are shaped by a multileaf collimator and modulated in intensity (IMRT, VMAT) to conform dose to the tumour while sparing normal tissue. The MV build-up region spares skin, and the dose is fractionated.

BME-374 · How does an oxygen concentrator work, and what are its design considerations?

Pressure swing adsorption: compressed air passes through zeolite beds that adsorb N₂, producing oxygen of roughly 90–95% purity. Two beds alternate between adsorption and purge. Design must address flow versus purity, noise, power, heat, filter maintenance and low-purity alarms.

Tissue Engineering

BME-375 · Estimate the maximum thickness of a cell-laden slab fed from one face, with D = 2×10⁻⁹ m²/s, surface O₂ = 0.2 mol/m³ and consumption Q = 0.02 mol/m³/s.

For zero-order consumption, L_max = √(2DC₀/Q) = √(2 × 2×10⁻⁹ × 0.2/0.02) = 2×10⁻⁴ m, or about 200 µm. This illustrates the diffusion limit.

BME-376 · How would you design a tissue-engineered graft for a critical-sized bone defect?

A porous, osteoconductive scaffold (β-TCP or HA composite, or 3D-printed PCL/TCP) with interconnected pores, seeded with autologous MSCs. Add controlled-release osteoinductive factors such as low-dose BMP-2, a vascularization strategy and rigid fixation. Validate in a large-animal model.

BME-377 · What are the main challenges in scaling up cell-based therapies?

Donor-to-donor variability. Expanding cells without losing phenotype. Closed, automated GMP processing. Potency and identity release assays. Sterility and mycoplasma testing. Cryopreservation and logistics. Cost of goods. Regulatory pathways (biologics or HCT/Ps in the US, ATMPs in the EU).

BME-378 · How would you engineer articular cartilage?

Chondrocytes or MSCs in a hydrogel (agarose, alginate or hyaluronic acid) with TGF-β3. Use dynamic compression or hydrostatic pressure bioreactors to build matrix. Mimic the zonal architecture and aim for integration with host cartilage. Measure compressive modulus, GAG and type II collagen against native tissue.

BME-379 · Calculate wall shear stress in a parallel-plate flow chamber with μ = 1 mPa·s, Q = 1 mL/min, w = 20 mm and h = 0.5 mm.

τ = 6μQ/(wh²). Q = 1.67×10⁻⁸ m³/s, so τ = 6 × 10⁻³ × 1.67×10⁻⁸ / (0.02 × 2.5×10⁻⁷) ≈ 0.02 Pa (0.2 dyn/cm²). This is low compared with arterial levels of about 1–2 Pa.

BME-380 · How does macrophage polarization affect scaffold outcomes?

M1 (pro-inflammatory) macrophages dominate early. A timely shift to M2 (pro-remodelling) phenotypes favours constructive remodelling rather than fibrosis. Scaffold chemistry, stiffness, degradation products, porosity and ECM derivation can be designed to promote M2 responses.

BME-381 · What are the design considerations for an organ-on-a-chip?

Physiological shear and mechanical strain, such as cyclic stretch for lung. Co-culture interfaces separated by porous membranes. Ratios of medium to tissue volume. Material choice: PDMS absorbs small hydrophobic drugs. Integrated sensors. Scaling rules when coupling multiple organs.

BME-382 · How do process parameters control fibre diameter in electrospinning?

Higher polymer concentration and viscosity increase diameter and prevent beading. Higher flow rate increases diameter. Voltage and tip-to-collector distance affect jet stretching and solvent evaporation. A rotating mandrel collector produces aligned fibres for tendon or nerve guides.

BME-383 · Why might PCL be chosen for a 3D-printed, resorbable airway splint for an infant?

PCL melts at about 60 °C, which makes FDM printing easy. It is ductile and degrades slowly, over about 2–3 years. That gives support long enough for the trachea to grow and remodel before resorption.

BME-384 · How do you characterize the quality of an engineered tissue?

Viability (live/dead staining), histology and immunostaining, biochemical content (DNA, GAG, collagen by hydroxyproline), gene expression, mechanical properties compared with native tissue, sterility and endotoxin testing, and functional performance in vivo.

Biosensors

BME-385 · What are the key elements of a subcutaneous CGM design?

A GOx electrode with a diffusion-limiting, biocompatible outer membrane. Interference rejection. Temperature compensation. Calibration (factory or user). Lag compensation and noise filtering. Resistance to biofouling and the foreign body response. Wireless telemetry. Accuracy reported as MARD.

BME-386 · Compute the MARD for CGM/reference pairs 110/100, 90/100 and 200/180 mg/dL.

Absolute relative differences are 10%, 10% and 11.1%. MARD = (10 + 10 + 11.1)/3 ≈ 10.4%.

BME-387 · Estimate the diffusion-limited current for n = 2, A = 1 mm², D = 1×10⁻¹¹ m²/s in the membrane, C = 5 mM and membrane thickness 10 µm.

i = nFADC/δ = 2 × 96485 × 10⁻⁶ × 10⁻¹¹ × 5 / 10⁻⁵ ≈ 0.96 µA. The membrane, not enzyme kinetics, then sets the sensitivity.

BME-388 · How do you reduce electroactive interference in amperometric glucose sensors?

Ascorbate, urate and acetaminophen oxidize at +0.6 V. Use permselective inner membranes (Nafion, cellulose acetate, polyphenylenediamine), mediators or wired enzymes that operate at low potential, or a sentinel electrode without enzyme for subtraction.

BME-389 · How can a lateral flow assay be made more sensitive?

Use brighter or larger labels (gold nanoshells, fluorescent or quantum-dot labels with a reader), silver or enzymatic amplification, optimized membrane flow rate and antibody density, sample pre-concentration, and quantitative optical readers.

BME-390 · An SPR assay gives ka = 1×10⁵ M⁻¹s⁻¹ and kd = 1×10⁻³ s⁻¹. What is K_D?

K_D = kd/ka = 10⁻³/10⁵ = 10⁻⁸ M, or 10 nM, which is a moderately high-affinity interaction.

BME-391 · What are the main challenges in wearable sweat biosensors?

Very small, variable sweat volumes. Concentrations that depend on sweat rate. Uncertain correlation with blood levels. Contamination from skin and old sweat. Evaporation. Sensor drift and calibration. Temperature and pH effects. Microfluidics and periodic induction (for example iontophoresis) help.

BME-392 · Why does Debye screening limit FET biosensors in physiological samples?

At physiological ionic strength (about 150 mM), the Debye length is about 0.7 nm. Charges on bound targets beyond this distance are screened, so a large antibody–antigen complex produces little signal. Remedies are dilution, small receptors such as aptamers or fragments, and polymer layers.

BME-393 · How do you manage drift in long-term implanted sensors?

Use stable reference electrodes and membranes, on-board self-checks such as impedance monitoring, recalibration, and model-based correction with Kalman filters. Fault detection should flag readings as unreliable rather than report them.

BME-394 · How would you validate a new blood glucose meter?

Run a clinical accuracy study against a laboratory reference following ISO 15197, using error-grid analysis. Add interference testing, haematocrit, temperature, humidity and altitude studies, lot-to-lot precision, and user-performance evaluation with lay users.

Regulatory Standards & Quality

BME-395 · What are the key regulatory standards for medical devices?

ISO 13485 (quality management), IEC 60601 series (electrical safety and essential performance), ISO 14971 (risk management), IEC 62304 (software lifecycle), IEC 62366-1 (usability), ISO 10993 (biocompatibility), plus market routes: FDA 510(k), De Novo or PMA, and EU MDR CE marking.

BME-396 · How would a standalone app that detects atrial fibrillation from ECG be classified?

Under EU MDR Rule 11, software providing information for diagnostic decisions is at least Class IIa. It rises to IIb if decisions could cause serious deterioration, and to III for death or irreversible harm. In the US, comparable ECG software functions are Class II.

BME-397 · A fault has a probability of 10⁻³ of creating a hazardous situation, and that situation leads to harm 10% of the time. What is the probability of harm?

Under ISO 14971, P(harm) = P₁ × P₂ = 10⁻³ × 0.1 = 10⁻⁴ per use. Combine this with severity in the risk matrix to judge acceptability.

BME-398 · What should design control traceability show?

User needs → design inputs → design outputs → verification evidence → validation evidence, with risk controls linked to requirements and tests. The traceability matrix demonstrates completeness and supports change impact analysis.

BME-399 · How should SOUP be handled under IEC 62304?

Identify each SOUP item and its version. Specify the functional, performance and hardware requirements it must meet. Evaluate known anomalies for safety impact. Include it in risk analysis, with segregation or mitigations. Control updates through configuration management.

BME-400 · What cybersecurity evidence is expected for connected devices?

Threat modelling and security risk assessment, secure design (authentication, encryption, least privilege), an SBOM, penetration testing, vulnerability management and patching plans, and coordinated disclosure. The FDA requires these for "cyber devices", and IEC 81001-5-1 defines secure lifecycle processes.

BME-401 · What does the EU MDR require for clinical evaluation?

A clinical evaluation plan and report based on literature, clinical investigations and PMCF data. Claims of equivalence need demonstrated technical, biological and clinical equivalence, plus data access for implantable and Class III devices. Clinical investigations follow ISO 14155.

BME-402 · When is process validation (IQ/OQ/PQ) required?

When the output of a process cannot be fully verified by subsequent inspection, as with sterilization, sealing, molding and welding. IQ confirms correct installation, OQ establishes operating limits and PQ shows consistent results under routine conditions.

BME-403 · With an initial bioburden of 100 CFU and D₁₂₁ = 1.5 min, how long is needed to reach SAL 10⁻⁶?

Required log reduction = log₁₀(100) + 6 = 8. Time = 8 × D = 8 × 1.5 = 12 min at 121 °C.

BME-404 · What is needed to bring a 510(k)-cleared device to the EU under the MDR?

Reclassify the device under Annex VIII. Prepare technical documentation (Annexes II and III) with a GSPR checklist, a clinical evaluation report, a PMS/PMCF plan, UDI and EUDAMED registration, and a person responsible for regulatory compliance. Most classes also need Notified Body assessment. FDA clearance is not accepted as equivalent.

BME-405 · What is FDA's Quality Management System Regulation (QMSR)?

The revision of 21 CFR Part 820 that incorporates ISO 13485:2016 by reference, harmonizing the US quality system requirements with the international standard. It took effect in February 2026, with some FDA-specific additions.

Clinical Engineering & Electrical Safety

BME-406 · Estimate the earth leakage caused by 4.7 nF of mains filter capacitance at 230 V, 50 Hz.

I = V × 2πfC = 230 × 2π × 50 × 4.7×10⁻⁹ ≈ 0.34 mA. Filter Y-capacitors are a major source of enclosure and earth leakage, so they must be sized against limits.

BME-407 · A patient has a central line, and two nearby devices' grounds differ by 50 mV. Is this a risk?

If a conductive catheter connects the heart to one device, a 50 mV difference over about 500 Ω drives about 100 µA, which is in the microshock VF range. Mitigate with equipotential bonding, CF-rated isolated equipment and insulated catheter connections.

BME-408 · How would you plan medical equipment replacement?

Score devices on age against expected life, end of manufacturer support, failure and downtime history, cost of service, safety alerts, utilization, clinical need and technological obsolescence. Prioritize within the capital budget, and include total cost of ownership and interoperability in the decision.

BME-409 · What should a clinical engineer include in a technology procurement evaluation?

Clinical requirements and specifications, regulatory status, total cost of ownership (consumables, service, training), interoperability and cybersecurity, usability trials with staff, vendor support and parts availability, and acceptance testing on delivery.

BME-410 · How can alarm fatigue in patient monitoring be reduced?

Customize thresholds and alarm delays for each unit, change electrodes daily and prepare skin, use smart or combined-parameter alarms, prioritize alarms clearly (IEC 60601-1-8), route them to the responsible caregiver, and audit alarm data regularly.

BME-411 · What elements make up an MRI safety program?

Four access-controlled zones, screening of patients and staff, ferromagnetic detection, and labelling of equipment as MR Safe, MR Conditional or MR Unsafe. Add conditional-implant protocols, SAR and burn precautions, quench and emergency procedures, and trained MR safety officers.

BME-412 · How do you mitigate surgical fire risk?

Manage the fire triangle. Oxidizer: minimize oxygen enrichment near the head and neck. Ignition: control ESU and laser tips, use holsters. Fuel: let alcohol preps dry, choose drapes carefully. Communicate fire risk at time-out and have extinguishing plans ready.

BME-413 · A Class I device develops a live-to-chassis fault. Why does a 0.1 Ω protective earth matter?

A low-impedance earth path draws a large fault current (hundreds of A with low supply impedance). This blows the fuse or trips the breaker within milliseconds, limiting the duration of chassis voltage. A high-resistance earth may keep the chassis live without tripping.

BME-414 · An infusion pump is suspected in a patient overdose. What do you do?

Remove it from service with its set and drug, untouched. Preserve event logs and the pump configuration. Document the scene. Involve risk management and the manufacturer. Analyse human factors as well as hardware. Report to regulators as required. Check the fleet for similar risk.

BME-415 · What does IEC 80001-1 address?

Risk management for IT networks that incorporate medical devices. It covers responsibilities of the healthcare organization, manufacturers and IT providers for safety, effectiveness and security, including change management, network segmentation and incident response.

Healthcare Informatics & AI

BME-416 · Outline the development of a deep-learning chest X-ray triage model.

Define the intended use. Curate multi-site DICOM data with radiologist-verified labels. Split by patient. Preprocess and augment the images, then fine-tune a pretrained CNN. Calibrate outputs, analyse subgroups, validate externally and evaluate clinically. Document the work under the QMS and the applicable SaMD pathway.

BME-417 · A validated imaging AI model performs poorly at a new hospital. Why, and what do you do?

Dataset shift: different scanners, protocols, populations, disease prevalence or labelling practices. Detect it with performance and input-distribution monitoring. Mitigate by local validation, recalibration, domain adaptation or retraining under controlled change procedures.

BME-418 · From TP = 80, FP = 20, FN = 40 and TN = 860, calculate precision, recall, F1 and specificity.

Precision = 80/100 = 0.80. Recall = 80/120 ≈ 0.67. F1 = 2PR/(P + R) ≈ 0.73. Specificity = 860/880 ≈ 0.98.

BME-419 · What is a Predetermined Change Control Plan for AI-enabled devices?

A plan submitted to and authorized by FDA that describes the specific planned model modifications, the protocol for developing and validating them, and an impact assessment. Changes made within the plan do not need a new marketing submission.

BME-420 · What are the limitations of saliency maps for clinical AI explainability?

Methods such as Grad-CAM show where the model attended, not why. They can be unstable and can highlight confounders, for example text markers. They do not prove correctness. They support, but do not replace, rigorous validation and human oversight.

BME-421 · How would you integrate bedside device data into an EHR using FHIR?

Collect device data through IEEE 11073 or vendor interfaces at a gateway. Map them to FHIR Observation and Device resources with LOINC or IEEE codes. Send them over a secure REST API with OAuth 2.0 (SMART on FHIR). Validate units, time synchronization and patient identification.

BME-422 · How can algorithmic bias arise in medical AI, and how do you address it?

Through under-represented groups in training data, biased labels or proxies (for example cost as a proxy for need), and sensor biases such as pulse oximetry in darker skin. Address it with diverse data, subgroup performance reporting, fairness metrics, bias audits and post-market monitoring.

BME-423 · Why does calibration matter in clinical risk models, and how is it assessed?

Decisions depend on absolute risk, so a well-ranked but miscalibrated model can mislead treatment thresholds. Assess calibration with calibration plots, calibration slope and intercept, and the Brier score. Correct it with Platt scaling, isotonic regression or recalibration on local data.

BME-424 · How does the EU AI Act affect AI medical devices?

AI systems that are medical devices, or safety components of them, requiring Notified Body assessment under the MDR or IVDR are classed as high-risk. They need risk management, data governance, technical documentation, logging, transparency, human oversight, accuracy and robustness, assessed alongside MDR conformity.

BME-425 · How can patient data be shared for analytics while preserving privacy?

Apply de-identification and pseudonymization with re-identification risk assessment (k-anonymity, for example). Add differential privacy for aggregate outputs, federated learning, secure enclaves or trusted research environments, data use agreements, and access controls with audit trails.

Other levels: Basic · Intermediate · Tricky · All Biomedical Engineering questions