Repository navigation
In this release: The ADMIRE model-based beamformer, the UltraPIPS perceptual metric for B-mode images, and improved streaming infrastructure for training on large ultrasound datasets through Hugging Face!
New Features
- ADMIRE (Aperture Domain Model Image REconstruction), a model-based beamformer that suppresses off-axis and multipath clutter, as
zea.ops.ADMIREorzea.ops.Beamform(beamformer="admire"). Seezea.beamform.admire(#650 by @tristan-deep) - UltraPIPS, a perceptual metric for B-mode ultrasound built on an ultrasound foundation model:
zea.models.ultrapips.UltraPIPS, or theultrapipsmetric inzea.metrics, used likelpips(#649 by @tristan-deep) - Transparent GIFs from RGBA frames with
transparent=Trueinzea.io_lib.save_videoandzea.io_lib.save_to_gif(#663 by @tristan-deep) - GIFs that play once, with
loop=Noneinzea.io_lib.save_to_gif(#680 by @wesselvannierop) zea.func.vmapaccepts nested inputs, dict axes and keyword arguments, likejax.vmap, and can be nested on all backends (#692 by @wesselvannierop)zea appcan serve over HTTPS with--ssl-certfileand--ssl-keyfile(#648 by @Louisvh)- Frequency-dependent attenuation in the pressure field, with
attenuation_coefin dB/(cm·MHz) inzea.beamform.pfield.compute_pfield. It defaults to 0 (lossless) (#682 by @wesselvannierop) zea.ops.BandPassFilterandzea.ops.LowPassFilterIQtake theirbandwidthfrom the newzea.Parameters.bandwidth, derived ascenter_frequency * probe_bandwidth_percent / 100unless set orcenter_frequencyvaries across transmits (#644 by @tristan-deep)
Changed
- Streaming
hf://datasets no longer runs into the Hugging Face rate limit, so you can train on many batches with many workers without downloading the data. The API is unchanged; two new environment variables,ZEA_MAX_CONCURRENT_REQUESTSandZEA_FIND_H5_SHAPES_WORKERS, tune it (see Environment variables) (#671 by @tristan-deep) - Conversion scripts for EchoNet-Dynamic, CAMUS and PICMUS accept the
.zipfile itself. EchoNet-Dynamic reproduces the published split by default, and its polar conversion is about 15 times faster (#676 by @wesselvannierop) - (Breaking) In
compute_pfield,alphaandpercentilealso shape the field whennorm=False, the unusedn_elargument is removed, and probes other than linear and phased arrays raiseNotImplementedError(#657 by @wesselvannierop) - (Breaking)
zea.func.vmapvalidates axes likejax.vmap, so a list no longer matches a tuple.out_axes=Noneonly works on the nativejax.vmap, and the_use_torch_vmapargument is removed (#692 by @wesselvannierop) Parameters.copy()andset_transmitsare much faster (#656 by @wesselvannierop)- (Breaking) The check on cartesian beamforming grids uses per-axis limits from the image's k-space support instead of λ/2, see How fine should the beamforming grid be?. Auto-sized grids get one more pixel per axis (#673 by @wesselvannierop)
- The default
probe_bandwidth_percentis 100 instead of 200, which also narrows the band used bycompute_pfield(#644 by @tristan-deep)
Fixed
- A streamed
hf://read that still fails after its retries, such as a dropped connection, no longer ends the epoch: theDataloaderreopens the file and reads the sample again (#690 by @tristan-deep) zea.func.vmappasses arguments at the right positions with multipleNonearguments. It also keeps all outputs, supports nested andNoneoutputs, supports negative axes when batching, and raises an error for arguments of different lengths (#691 by @wesselvannierop)- The pressure field uses the one-way transmit spectrum, averages over frequencies instead of summing, and takes the element positions from
probe_geometry(#657 by @wesselvannierop) t_peakestimated fromwaveforms_two_wayuses the envelope peak; values change where the carrier peak differs from it (#656 by @wesselvannierop)- Conversion of Verasonics
BS100BWandBS50BWdata samples I and Q at the same time instants. Data converted with earlier versions, such as some Open-H-RF data, may be affected (#672 by @vincentvdschaft) zea.ops.Normalizeclips its output instead of its input, so results stay withinoutput_range(#681 by @wesselvannierop)waveforms_one_wayandwaveforms_two_wayfollowselected_transmits, like the other per-transmit parameters (#683 by @wesselvannierop)- Operations and pipelines can be deep-copied and pickled, also when not registered with Keras, and copies keep their JIT settings (#698 by @tristan-deep)
- The shared GIF palette keeps exact colors, so pure black stays
(0, 0, 0)(#680 by @wesselvannierop) - Deriving
xlimson a GPU with the torch backend (#661 by @tristan-deep) - Jitted ops on the torch backend in the Docker images, which lacked a C++ compiler (#658 by @wesselvannierop)
zea appandload_model_from_hfdownload to the zea cache instead of the Hugging Face cache (#689 by @Louisvh)@cache_outputfunctions that call a recursive function can now be cached. Additionally, the cache key now also includes other@cache_outputfunctions it calls, so editing those invalidates its cache (#703 by @wesselvannierop)
Full Changelog: v0.1.8...v0.1.9