Skip to content
Murali RaviMedical Instrumentation

Research

Four instruments,
documented in full.

Each work has a dedicated case study: hero specifications, methods, development timeline, outcomes, collaborators and purpose-built figures.

Works
4
Span
2012-2026
Figures
16
Clinically validated
2
CONVENTIONAL: INDIRECTγCollimatorScintillator (NaI / LaBr₃)Light guidePhotomultiplier arrayReadout electronics~204 mmDETECTOR DEPTHSAIGC: DIRECT CONVERSIONγLEHR collimatorCZT: direct conversionASIC readoute⁻ charge~110 mmDETECTOR DEPTHWHY IT MATTERSNo light stage. No PMT depth.Room temperature · portable form factorCZT SELECTED AFTER GEANT4 / GATE MONTE CARLO COMPARISON AGAINST LaBr₃ SCINTILLATORS
Clinically validated2016-2021

Small Area Imaging Gamma Camera

A portable, high-resolution nuclear imaging device: conceived, built and taken to the clinic.

100%correlation with gold-standard LFOV cameras

88 × 88 mm

Central field of view

1,936

Detector elements

2 mm

Intrinsic spatial resolution

Open case study
SURGICAL FIELDTUMOUR MARGINSENTINEL NODEOPTICAL780 nmGAMMA140 keVSAIGC-TORCHICG FLUORESCENCEWhere does the tumour end?⁹⁹ᵐTc GAMMAWhere has it spread?COMPLEMENTARY, NOT REDUNDANT: FLUORESCENCE READS THE VISIBLE SURFACE MARGIN; GAMMA REACHES DEPTH AND METABOLIC ACTIVITY.CHANNEL 01: OPTICALIndocyanine GreenExcited at 780 nm · high sensitivityCHANNEL 02: NUCLEARTechnetium-99m140 keV · nodal & metabolic
Clinically validated2017-2023

Dual-Modality Intraoperative Cancer Navigation

A first-of-its-kind handheld probe that lets a surgeon confirm cancer while the patient is still on the table.

100%successful detection in oral, throat and breast cancer trials

780 nm

ICG fluorescence excitation

140 keV

Technetium-99m gamma detection

0 days

Wait for validation (was 3-4)

Open case study
MAXIMUM LIKELIHOOD EXPECTATION MAXIMIZATION: ITERATIVE LOOPForwardprojection1Compareto measured2Backprojection3Updateestimate4× N iterationsUNTIL CONVERGENCETHE UPSIDEHighest imagequality availableNoise modelling, better contrastTHE COSTExecution timeblocks clinical useEvery iteration re-projectsThe trade-off is not a law of nature: it is an architecture problem.
Deployed2016-2019

Hardware Acceleration for Medical Imaging

Making the best reconstruction algorithm fast enough to actually use.

288×MLEM speed-up on FPGA over optimised Xeon software

288×

FPGA acceleration (MLEM)

250%

DSP processing acceleration

Virtex-7

Xilinx VC709 platform

Open case study
ECAPA-LITE: 128-CHANNEL LIGHTWEIGHT ECAPA-TDNN VARIANTWaveform16 kHz audioMFCCOn-chipConv1D128 chSE-Res2Block×3, dilatedAttentive StatPoolingEmbeddingSpeaker vectorCHANNEL REDUCTION: WHY LITEECAPA-TDNN512 channelsECAPA-LITE128 channelsSELECTED FOR ACCURACY-TO-SIZE RATIO: THE METRIC THAT MATTERS WHEN DEPLOYMENT IS MEASURED IN MEGABYTESEQUAL ERROR RATE2.17%VoxCeleb1 test set, full precisionCHANNELS128Lightweight variant widthQUANTISED SIZE1.57 MiBINT8 post-training
Proof of concept2025-2026

Edge-Deployable Voice Biometric Authentication

Speaker verification at 2.17% EER, compressed to 1.57 MiB, running on the device itself.

2.17% EERon the VoxCeleb1 test set

2.17%

Equal Error Rate (float)

2.60%

EER after INT8 quantisation

1.57 MiB

Quantised model size

Open case study