Add (multi) 2D EPI kernel and scripts by schuenke · Pull Request #70 · PTB-MR/mrseq

schuenke · 2026-02-25T11:47:36Z

To do:

verify mrd trajectory information for multi-2D acquisitions with and without noise and/or navigator acquisitions
add tests
add example scripts
(optional) replace pulseq-based traj calculation with analytical definition (including ramp sampling and oversampling)
add frequency offset for ADC in EPI readout

github-actions · 2026-02-25T11:50:13Z

Coverage Report

File	Stmts	Miss	Cover	Missing
/home/runner/.local/lib/python3.12/site-packages/mrseq
_version.py	7	7	0%	2–9
/home/runner/.local/lib/python3.12/site-packages/mrseq/preparations
t2_prep.py	51	1	98%	195
/home/runner/.local/lib/python3.12/site-packages/mrseq/scripts
epi2d_fid.py	155	9	94%	308, 411, 468–473, 485, 491
epi2d_se.py	173	14	92%	205–207, 330, 358, 366, 469, 526–531, 543, 549
grpe_flash_dixon.py	205	18	91%	121, 124, 189, 206–208, 330, 349, 400, 410, 503, 571–576, 596, 602
radial_flash.py	140	11	92%	101, 104, 259, 345, 403–408, 428, 434
spiral_flash.py	120	10	92%	95, 231, 313, 363–368, 387, 393
t1_inv_rec_gre_single_line.py	93	9	90%	162, 255, 298–303, 330, 336
t1_inv_rec_se_single_line.py	106	9	92%	219, 316, 375–380, 407, 413
t1_molli_bssfp.py	141	13	91%	216, 282–284, 299, 374, 428–433, 454, 460
t1_t2_spiral_cmrf.py	149	12	92%	89, 95, 114, 168, 385, 432–437, 463, 469
t1rho_se_single_line.py	102	8	92%	310, 367–372, 399, 405
t2_multi_echo_se_single_line.py	96	9	91%	162, 261, 314–319, 344, 350
t2_t2prep_flash.py	143	15	90%	103, 226, 243, 277–279, 291, 367, 429–434, 454, 460
t2star_multi_echo_flash.py	145	14	90%	114, 233, 280–282, 294, 390, 452–457, 477, 483
/home/runner/.local/lib/python3.12/site-packages/mrseq/utils
EpiReadout.py	306	90	71%	52, 61, 193, 197, 244, 318, 322–333, 455, 508–592, 658–664, 684–729
ismrmrd.py	93	2	98%	173, 179
sequence_helper.py	59	4	93%	105, 133–134, 162
trajectory.py	148	3	98%	45, 87, 358
TOTAL	2564	258	90%

Tests	Skipped	Failures	Errors	Time
1253	0 💤	4 ❌	0 🔥	2m 22s ⏱️

…_trajectory

schuenke · 2026-03-04T09:29:53Z

This is how the current version of plot_trajectory() looks like:

symmetric:

flyback:

ckolbPTB · 2026-03-20T19:07:12Z

In dc500da I changed how the amplitude of the readout gradient is modified for the symmetric EPI readout. In the previous version the EPI-object was modified. This meant that for an odd number of phase encoding steps, the readout gradient of the EPI-object was flipped at the end of add_to_seq. The EPI-object could then not be used anymore to obtain e.g. another repetition. Now the EPI-object is not modified.

I removed the rev-label here:

mrseq/src/mrseq/utils/EpiReadout.py

Line 633 in dc500da

seq.add_block(self.gx, self.adc, lin_label)

I think this should be fine but maybe @schuenke could double check

ckolbPTB · 2026-04-30T11:36:32Z

+                    epi2d.adc,
+                    pp.make_label(label='REV', type='SET', value=gx.amplitude < 0),
+                    pp.make_label(label='SEG', type='SET', value=gx.amplitude < 0),
+                    pp.make_label(label='AVG', type='SET', value=(n + 1) == 3),


Suggested change

pp.make_label(label='AVG', type='SET', value=(n + 1) == 3),

I would remove this, because it leads to 1 readout having avg = 0 and two readouts with avg = 0,1 which then leads to problems for the MRpro sorting routines. They are not really averages anyway because there is T2* decay.

ckolbPTB · 2026-04-30T11:38:05Z

+                    epi2d.adc,
+                    pp.make_label(label='REV', type='SET', value=gx_sign < 0),
+                    pp.make_label(label='SEG', type='SET', value=gx_sign < 0),
+                    pp.make_label(label='AVG', type='SET', value=(n + 1) == 3),


Suggested change

pp.make_label(label='AVG', type='SET', value=(n + 1) == 3),

same as for epi2d_fid

ckolbPTB · 2026-04-30T11:40:36Z

+            )
+
+            # Navigator kx offset: starts from -gx_pre.area (reversed pre-winder)
+            nav_kx_offset = -epi2d.gx_pre.area


Suggested change

nav_kx_offset = -epi2d.gx_pre.area

nav_kx_offset = epi2d.gx_pre.area

I had to make these changes to get the correct navigator trajectory

ckolbPTB · 2026-04-30T11:40:45Z

+
+            # Navigator kx offset: starts from -gx_pre.area (reversed pre-winder)
+            nav_kx_offset = -epi2d.gx_pre.area
+            nav_gx_sign = -1.0  # first navigator uses reversed gx


Suggested change

nav_gx_sign = -1.0 # first navigator uses reversed gx

I had to make these changes to get the correct navigator trajectory

ckolbPTB · 2026-04-30T11:41:54Z

+                    assert prot is not None
+                    n_samples = epi2d.adc.num_samples
+                    traj = np.zeros((n_samples, 2), dtype=np.float32)
+                    if nav_gx_sign > 0:


Suggested change

if nav_gx_sign > 0:

ckolbPTB · 2026-04-30T11:42:04Z

+                    traj = np.zeros((n_samples, 2), dtype=np.float32)
+                    if nav_gx_sign > 0:
+                        nav_kx = nav_kx_offset + nav_kx_forward
+                    else:


Suggested change

else:

ckolbPTB · 2026-04-30T11:42:12Z

+                    if nav_gx_sign > 0:
+                        nav_kx = nav_kx_offset + nav_kx_forward
+                    else:
+                        nav_kx = nav_kx_offset - nav_kx_forward


Suggested change

nav_kx = nav_kx_offset - nav_kx_forward

ckolbPTB · 2026-04-30T11:42:36Z

+                    n_samples = epi2d.adc.num_samples
+                    traj = np.zeros((n_samples, 2), dtype=np.float32)
+                    if nav_gx_sign > 0:
+                        nav_kx = nav_kx_offset + nav_kx_forward


Suggested change

nav_kx = nav_kx_offset + nav_kx_forward

nav_kx = nav_kx_offset + gx_sign * nav_kx_forward

ckolbPTB · 2026-04-30T11:45:38Z

The above changes were necessary to get the correct trajectory for the navigators. Probably similary changes are needed for the epi2d_fid.

We could add the part on reading out the navigators to the example notebook. What do you think @schuenke ?

schuenke added 10 commits November 28, 2025 09:30

add EpiReadout (work in progress)

148ef91

ensure round_to_raster returns floats

a7f092c

update EpiReadout and add epi script (WIP)

9b7a9fa

add segmentation (SEG) label for reverse acquisitions

ec7e3dc

update epi2d script (deprecated)

6ae5063

split epi2d script into dedicated epi2d_fid and epi2d_se scripts (WIP)

005ec8d

minor adjustments to epi2d_se

406e7a8

Merge branch 'main' into epi2d

e33f0e8

cleanup docstrings

7459abe

fix mrd trajectory definitions for multi-slice

567cc09

ckolbPTB added 4 commits February 28, 2026 19:49

Merge branch 'main' into epi2d

b085211

tests added

0de0807

tests and example added. ismrmrd header fixed

1ebea4d

separate epi readout + tests

81338c8

schuenke mentioned this pull request Mar 3, 2026

Allow nufft for traj where nufft/fft splitting fails PTB-MR/mrpro#972

Merged

schuenke and others added 10 commits March 3, 2026 11:30

replace traj calculation for mrd file with analytical expression

2543378

remove rounding of traj values in mrd file

e744aee

add tests for analytical epi traj

2c1d2d4

fix mypy errors

8403cb3

add direct recon for all

a53c9b0

fix se epi2d

01f9dc5

update

6ed541b

adjust for v141 compatibility

468cac4

add calculate_full_trajectory method for better visualization in plot…

7e4cd03

…_trajectory

fix mypy error

74f42be

schuenke and others added 3 commits March 5, 2026 15:57

Merge branch 'main' into epi2d

f2b5b2d

Merge branch 'main' into epi2d

2e1ed14

Merge branch 'main' into epi2d

f4bc3ad

ensures epi object stays unchanged

dc500da

Merge branch 'main' into epi2d

123588c

ckolbPTB reviewed Apr 30, 2026

View reviewed changes

	nav_kx_offset = -epi2d.gx_pre.area
	nav_kx_offset = epi2d.gx_pre.area

	nav_kx = nav_kx_offset + nav_kx_forward
	nav_kx = nav_kx_offset + gx_sign * nav_kx_forward

Conversation

schuenke commented Feb 25, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

github-actions Bot commented Feb 25, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

schuenke commented Mar 4, 2026

Uh oh!

ckolbPTB commented Mar 20, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

ckolbPTB commented Apr 30, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

schuenke commented Feb 25, 2026 •

edited

Loading

github-actions Bot commented Feb 25, 2026 •

edited

Loading