From d313c54613ea12fd07a6bd91fc613b1f9d97288e Mon Sep 17 00:00:00 2001 From: Mathieu Boudreau Date: Wed, 7 Feb 2024 22:38:15 -0400 Subject: [PATCH] Update authors --- content/index.ipynb | 94 +++++++++++++-------------------------------- paper.md | 19 ++++----- 2 files changed, 37 insertions(+), 76 deletions(-) diff --git a/content/index.ipynb b/content/index.ipynb index 5d88a77..fb4ac6c 100644 --- a/content/index.ipynb +++ b/content/index.ipynb @@ -2,7 +2,7 @@ "cells": [ { "cell_type": "code", - "execution_count": 1, + "execution_count": null, "metadata": { "tags": [ "remove_output", @@ -48,7 +48,7 @@ "\n", "\n", "

\n", - "Daniel Papp1, Kyle M. Gilbert2,3, Gaspard Cereza1, Alexandre D’Astous1, Mathieu Boudreau1, Marcus Couch4, Pedram Yazdanbakhsh5, Robert L. Barry6,7,8, Eva Alonso Ortiz1, Julien Cohen-Adad1,9,10\n", + "Daniel Papp1, Kyle M. Gilbert2,3, Gaspard Cereza1, Alexandre D’Astous1, Nibardo Lopez-Rios1, Mathieu Boudreau1, Marcus Couch4, Pedram Yazdanbakhsh5, Robert L. Barry6,7,8, Eva Alonso Ortiz1, Julien Cohen-Adad1,9,10\n", "

\n", "\n", "\n", @@ -244,7 +244,7 @@ }, { "cell_type": "code", - "execution_count": 2, + "execution_count": null, "metadata": { "tags": [ "remove_output", @@ -277,22 +277,14 @@ }, { "cell_type": "code", - "execution_count": 3, + "execution_count": null, "metadata": { "tags": [ "remove_input", "remove_output" ] }, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "/Users/mathieuboudreau/neuropoly/github/rf-shimming-7t-neurolibre/content\n" - ] - } - ], + "outputs": [], "source": [ "# Start timer\n", "start_time = datetime.now()\n", @@ -303,7 +295,7 @@ }, { "cell_type": "code", - "execution_count": 4, + "execution_count": null, "metadata": { "tags": [ "remove_input", @@ -332,7 +324,7 @@ }, { "cell_type": "code", - "execution_count": 5, + "execution_count": null, "metadata": { "tags": [ "remove_input", @@ -382,24 +374,14 @@ }, { "cell_type": "code", - "execution_count": 6, + "execution_count": null, "metadata": { "tags": [ "hide_input", "remove_output" ] }, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "path_data: /Users/mathieuboudreau/neuropoly/github/rf-shimming-7t-neurolibre/data/rf-shimming-7t/ds004906\n", - "shim_modes: ['CP', 'patient', 'volume', 'phase', 'CoV', 'target', 'SAReff']\n", - "subjects: ['sub-01', 'sub-02', 'sub-03', 'sub-04', 'sub-05']\n" - ] - } - ], + "outputs": [], "source": [ "path_data = os.getcwd()\n", "print(f\"path_data: {path_data}\")\n", @@ -433,7 +415,7 @@ }, { "cell_type": "code", - "execution_count": 7, + "execution_count": null, "metadata": { "tags": [ "hide_input" @@ -470,7 +452,7 @@ }, { "cell_type": "code", - "execution_count": 8, + "execution_count": null, "metadata": { "tags": [ "hide_input" @@ -502,7 +484,7 @@ }, { "cell_type": "code", - "execution_count": 9, + "execution_count": null, "metadata": { "tags": [ "hide_input" @@ -527,7 +509,7 @@ }, { "cell_type": "code", - "execution_count": 10, + "execution_count": null, "metadata": { "tags": [ "hide_input" @@ -555,7 +537,7 @@ }, { "cell_type": "code", - "execution_count": 11, + "execution_count": null, "metadata": { "tags": [ "hide_input" @@ -581,7 +563,7 @@ }, { "cell_type": "code", - "execution_count": 12, + "execution_count": null, "metadata": { "tags": [ "hide_input" @@ -614,42 +596,17 @@ }, { "cell_type": "code", - "execution_count": 13, + "execution_count": null, "metadata": { "tags": [ "hide_input", "remove_output" ] }, - "outputs": [ - { - "ename": "FileNotFoundError", - "evalue": "[Errno 2] No such file or directory: '/Users/mathieuboudreau/neuropoly/github/rf-shimming-7t-neurolibre/data/rf-shimming-7t/ds004906/derivatives/results/sub-01_acq-CP_T2starw_label-SC.csv'", - "output_type": "error", - "traceback": [ - "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m", - "\u001b[0;31mFileNotFoundError\u001b[0m Traceback (most recent call last)", - "\u001b[0;32m/var/folders/l1/1xswjffd73l8yp7dd7pq9lyw0000gn/T/ipykernel_3003/2567902214.py\u001b[0m in \u001b[0;36m\u001b[0;34m\u001b[0m\n\u001b[1;32m 54\u001b[0m \u001b[0;31m# Get signal in SC\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 55\u001b[0m \u001b[0mfile_csv\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mos\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpath\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mjoin\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mpath_results\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34mf\"{subject}_acq-{shim_mode}_T2starw_label-SC.csv\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 56\u001b[0;31m \u001b[0mdf\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mpd\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mread_csv\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfile_csv\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m 57\u001b[0m \u001b[0mdata_sc\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mdf\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m'WA()'\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 58\u001b[0m \u001b[0mdata_sc_smoothed\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0msmooth_data\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdata_sc\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n", - "\u001b[0;32m~/opt/anaconda3/lib/python3.9/site-packages/pandas/io/parsers/readers.py\u001b[0m in \u001b[0;36mread_csv\u001b[0;34m(filepath_or_buffer, sep, delimiter, header, names, index_col, usecols, dtype, engine, converters, true_values, false_values, skipinitialspace, skiprows, skipfooter, nrows, na_values, keep_default_na, na_filter, verbose, skip_blank_lines, parse_dates, infer_datetime_format, keep_date_col, date_parser, date_format, dayfirst, cache_dates, iterator, chunksize, compression, thousands, decimal, lineterminator, quotechar, quoting, doublequote, escapechar, comment, encoding, encoding_errors, dialect, on_bad_lines, delim_whitespace, low_memory, memory_map, float_precision, storage_options, dtype_backend)\u001b[0m\n\u001b[1;32m 910\u001b[0m \u001b[0mkwds\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mupdate\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mkwds_defaults\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 911\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 912\u001b[0;31m \u001b[0;32mreturn\u001b[0m \u001b[0m_read\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfilepath_or_buffer\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mkwds\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m 913\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 914\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n", - "\u001b[0;32m~/opt/anaconda3/lib/python3.9/site-packages/pandas/io/parsers/readers.py\u001b[0m in \u001b[0;36m_read\u001b[0;34m(filepath_or_buffer, kwds)\u001b[0m\n\u001b[1;32m 575\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 576\u001b[0m \u001b[0;31m# Create the parser.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 577\u001b[0;31m \u001b[0mparser\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mTextFileReader\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfilepath_or_buffer\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwds\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m 578\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 579\u001b[0m \u001b[0;32mif\u001b[0m \u001b[0mchunksize\u001b[0m \u001b[0;32mor\u001b[0m \u001b[0miterator\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n", - "\u001b[0;32m~/opt/anaconda3/lib/python3.9/site-packages/pandas/io/parsers/readers.py\u001b[0m in \u001b[0;36m__init__\u001b[0;34m(self, f, engine, **kwds)\u001b[0m\n\u001b[1;32m 1405\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 1406\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mhandles\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0mIOHandles\u001b[0m \u001b[0;34m|\u001b[0m \u001b[0;32mNone\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1407\u001b[0;31m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_engine\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_make_engine\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mf\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mengine\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m 1408\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 1409\u001b[0m \u001b[0;32mdef\u001b[0m \u001b[0mclose\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m->\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n", - "\u001b[0;32m~/opt/anaconda3/lib/python3.9/site-packages/pandas/io/parsers/readers.py\u001b[0m in \u001b[0;36m_make_engine\u001b[0;34m(self, f, engine)\u001b[0m\n\u001b[1;32m 1659\u001b[0m \u001b[0;32mif\u001b[0m \u001b[0;34m\"b\"\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mmode\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 1660\u001b[0m \u001b[0mmode\u001b[0m \u001b[0;34m+=\u001b[0m \u001b[0;34m\"b\"\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1661\u001b[0;31m self.handles = get_handle(\n\u001b[0m\u001b[1;32m 1662\u001b[0m \u001b[0mf\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 1663\u001b[0m \u001b[0mmode\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n", - "\u001b[0;32m~/opt/anaconda3/lib/python3.9/site-packages/pandas/io/common.py\u001b[0m in \u001b[0;36mget_handle\u001b[0;34m(path_or_buf, mode, encoding, compression, memory_map, is_text, errors, storage_options)\u001b[0m\n\u001b[1;32m 857\u001b[0m \u001b[0;32mif\u001b[0m \u001b[0mioargs\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mencoding\u001b[0m \u001b[0;32mand\u001b[0m \u001b[0;34m\"b\"\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mioargs\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmode\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 858\u001b[0m \u001b[0;31m# Encoding\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 859\u001b[0;31m handle = open(\n\u001b[0m\u001b[1;32m 860\u001b[0m \u001b[0mhandle\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 861\u001b[0m \u001b[0mioargs\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmode\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n", - "\u001b[0;31mFileNotFoundError\u001b[0m: [Errno 2] No such file or directory: '/Users/mathieuboudreau/neuropoly/github/rf-shimming-7t-neurolibre/data/rf-shimming-7t/ds004906/derivatives/results/sub-01_acq-CP_T2starw_label-SC.csv'" - ] - }, - { - "data": { - "image/png": "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", - "text/plain": [ - "
" - ] - }, - "metadata": {}, - "output_type": "display_data" - } - ], + "outputs": [], "source": [ + "%matplotlib inline\n", + "\n", "# Make figure of CSF/SC signal ratio from T2starw scan\n", "\n", "# Go back to root data folder\n", @@ -1119,6 +1076,7 @@ "# Adjust the layout so labels and titles do not overlap\n", "plt.tight_layout()\n", "plt.savefig(os.path.join(path_results, 'fig_b1plus.png'), dpi=300, format='png')\n", + "\n", "plt.show()" ] }, @@ -1261,6 +1219,7 @@ }, "outputs": [], "source": [ + "%matplotlib inline\n", "# Select subject to show\n", "subject = 'sub-05'\n", "os.chdir(os.path.join(path_data, subject, \"fmap\"))\n", @@ -1376,6 +1335,7 @@ }, "outputs": [], "source": [ + "%matplotlib inline\n", "# PYTHON CODE\n", "# Module imports\n", "\n", @@ -1644,8 +1604,8 @@ " yref='paper'\n", " ),\n", ")\n", - "#iplot(fig, filename = 'figure4a', config = config)\n", - "plot(fig, filename = 'figure1.html', config = config)\n" + "#iplot(fig, filename = 'figure3', config = config)\n", + "plot(fig, filename = 'figure3.html', config = config)\n" ] }, { @@ -1659,7 +1619,7 @@ }, "outputs": [], "source": [ - "display(HTML('figure1.html'))" + "display(HTML('figure3.html'))" ] }, { @@ -1726,9 +1686,9 @@ "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", - "version": "3.9.13" + "version": "3.8.18" } }, "nbformat": 4, - "nbformat_minor": 2 + "nbformat_minor": 4 } diff --git a/paper.md b/paper.md index 8782a0a..16df509 100644 --- a/paper.md +++ b/paper.md @@ -9,31 +9,32 @@ tags: - 7T authors: - name: Daniel Papp - orcid: + orcid: 0000-0003-1481-1413 affiliation: 1 - name: Kyle M. Gilbert - orcid: + orcid: 0000-0003-3026-5686 affiliation: "2, 3" - name: Gaspard Cereza - orcid: affiliation: 1 - name: Alexandre D’Astous - orcid: + orcid: 0000-0003-0381-7334 + affiliation: 1 + - name: Nibardo Lopez-Rios + orcid: 0000-0002-4791-8260 affiliation: 1 - name: Mathieu Boudreau - orcid: + orcid: 0000-0002-7726-4456 affiliation: 1 - name: Marcus Couch - orcid: + orcid: 0000-0002-2732-8941 affiliation: 4 - name: Pedram Yazdanbakhsh - orcid: + orcid: 0000-0003-4456-5997 affiliation: 5 - name: Robert L. Barry - orcid: affiliation: "6, 7, 8" - name: Eva Alonso Ortiz - orcid: + orcid: 0000-0001-6590-7234 affiliation: 1 - name: Julien Cohen-Adad orcid: 0000-0003-3662-9532